Treatments for preventing recurrence of infection with Pseudomonas aeruginosa in people with cystic fibrosis

Abstract

Background

Chronic infection with Pseudomonas aeruginosa (PA) in cystic fibrosis (CF) is a source of much morbidity and mortality. Eradication of early PA infection is possible, but can recur in many individuals. We sought to examine strategies to delay the time to PA recurrence in people with CF.

Objectives

To establish whether secondary prevention strategies, using antibiotics or other therapies, increase the chances of people with CF remaining free from PA infection following successful eradication therapy.

Search methods

We searched the Cochrane Cystic Fibrosis Trials Register, compiled from electronic database searches and handsearching of journals and conference abstract books. We also searched ongoing trials registries and the reference lists of relevant articles and reviews.

Date of last search: 21 August 2019.

Selection criteria

Randomised controlled trials (and quasi‐randomised trials where the risk of bias was low) comparing any treatment modality aimed at preventing recurrence of PA infection with placebo, standard therapy or any other treatment modality in people with CF who have undergone successful eradication of PA.

Data collection and analysis

Two review authors independently assessed trials for inclusion and risk of bias. Quality of the evidence was assessed using GRADE. Conflicts were resolved by discussion and the opinion of a third review author was sought where necessary. Only a subset of participants in the included trial were eligible, therefore individual participant data were requested and obtained from the trial investigators.

Main results

We included one trial (n = 306) in the review; however, only 253 participants had undergone successful eradication of PA, so fulfilling the inclusion criteria for our review. Information presented relates only to the included subset of participants. The trial recruited children aged one to 12 years (mean (standard deviation (SD)) age of 5.8 (3.5) years), 129 participants (51.0%) were female and the median follow‐up was 494 days. We compared cycled therapy with tobramycin inhalation solution (TIS), in which participants underwent 28 days of TIS every three months, with culture‐based therapy, in which participants were only prescribed medication when a quarterly sputum sample was positive for PA. Reasons for downgrading the quality of the evidence included applicability (only included children), incomplete outcome data and a small number of participants.

The time to next isolation of PA was probably shorter with cycled TIS therapy than with culture‐based therapy, hazard ratio (HR) 2.04 days (95% confidence interval (CI) 1.28 to 3.26) (moderate‐quality evidence). This is in contrast to the main publication of the only included trial, which examined rate of PA positivity rather than time to PA infection and included participants not eligible for inclusion in this review. At the end of the trial, there was no difference between the cycled and culture‐based groups in the change from baseline in forced expiratory volume in one second (FEV₁) L, mean difference (MD) 0.0 L (95% CI ‐0.09 to 0.09) or in FEV₁ % predicted, MD 0.70% (95% CI ‐4.33 to 5.73) (both very low‐quality evidence). There was no difference in the change from baseline for FVC between the groups. There was also no difference in the frequency of pulmonary exacerbations between groups, MD ‐0.18 (95% CI ‐0.51 to 0.14) (moderate‐quality evidence). Similarly, there was no difference between groups in the risk of participants developing novel resistant bacteria, RR 1.00 (95% CI 0.67 to 1.5) (moderate‐quality evidence). There were more severe adverse events in the cycled group, but the type of treatment probably makes little or no difference to the results, RR 0.65 (95% CI 0.39 to 1.11) (moderate‐quality evidence).

There was no difference between groups in the change in weight or height from baseline or in rates of adherence to tobramycin or all trial medicines. The included trial did not assess changes in quality of life, the time to chronic infection with PA or the cost‐effectiveness of treatment.

Authors' conclusions

Cycled TIS therapy may be beneficial in prolonging the time to recurrence of PA after successful eradication, but further trials are required, specifically addressing this question and in both adults and children.

Plain language summary

Treatments for preventing recurrence of infection with Pseudomonas aeruginosa in people with cystic fibrosis

Review question

We reviewed evidence about the effect of treatments given to people with cystic fibrosis (CF) to prevent recurrence of infection with bacteria called Pseudomonas aeruginosa (PA) after it has been successfully treated.

Background

People with CF experience frequent, severe chest infections. These infections may be caused by bacteria that do not cause disease in healthy people, such as PA. PA is an important infection in CF since if it is not treated early it cannot be cleared from the lungs (termed chronic PA). When it cannot be cleared from the lungs, it causes ongoing damage to the lung tissue as well as worse chest infections. People with CF who have chronic PA may be more unwell than those without it.

Early treatment can remove PA from the lungs of people with CF, but infection can happen again and is difficult to prevent. We want to know if giving extra treatment after the PA has been successfully treated, can prolong the time to the next infection. The treatment could be antibiotics or another treatment, for example something that helps the person's immune system fight the PA.

Search date

The evidence is current to: 21 August 2019.

Trial characteristics

We included one trial with 306 participants. Only 253 of these people had had successful treatment of PA and so could be included in our review. There were only children in the trial who were aged between one and 12 years old; 51% were girls. The trial compared treatment with one month of an inhaled antibiotic every three months ('cycled treatment') with antibiotic treatment given only when the person was found to have PA ('culture‐based treatment'). People were selected randomly to have either cycled or culture‐based treatment. The trial followed people up for an average of 70 weeks.

Key results

Taking into the account the certainty of the evidence, we believe that the time to the next isolation of PA was probably shorter with cycled therapy than with culture‐based therapy. We found that PA recurred in a quarter of participants in the culture‐based group by 249 days but in the cycled group it took 505 days; giving cycled treatment doubled the time to a new infection with PA compared with culture‐based treatment. This is in contrast to the main publication of the only included trial, which looked at the rate of positive PA cultures rather than the time to a new infection of PA and included participants not eligible for inclusion in this review. There was no difference between the groups in: pulmonary function (a measure of how well someone's lungs are working); in the number of people having chest infections; in the change in height and weight from the beginning to th end of the trial; in how many people took all the doses of the medications; in how many people developed infections with new bacteria; or in how many people had a serious complication. The trial did not give information about the effect of the two treatments on individuals' quality of life, the time until the development of chronic PA or the cost‐effectiveness of the treatment.

Quality of the evidence

The quality of most of the evidence was moderate. As the trial only included children, we cannot be sure if the cycled treatment would have the same effect in teenagers or adults with CF. Further trials including both adults and children are needed to help answer the question. Trials designed to specifically answer this question are needed.

Summary of findings

Summary of findings for the main comparison. Summary of findings: cycled versus culture‐based antibiotic therapy for preventing recurrence of infection with Pseudomonas aeruginosa.

Cycled antibiotic therapy compared with culture based antibiotic therapy for preventing recurrence of infection with PA in cystic fibrosis
Patient or population: children (aged 1 to 12 years) with CF and a newly isolated PA infection Settings: outpatients Intervention: cycled administration of antibiotics Comparison: culture‐based administration of antibiotics
Outcomes	Illustrative comparative risks* (95% CI)		Relative effect (95% CI)	No of participants (studies)	Quality of the evidence (GRADE)	Comments
	Assumed risk	Corresponding risk
	Culture‐based therapy	Cycled therapy
Time to next isolation of PA Follow‐up: up to 583 days	The participants in the culture‐based therapy group were twice as likely to have experienced a recurrence by their final study visit than those in the cycled group.		HR 2.04 (1.28 to 3.26)	253 (1)	⊕⊕⊕⊝ moderate1	HR was calculated from individual patient data.
Quality of life	This outcome was not measured.
FEV1: change from baseline (L) Follow‐up: median 494 days	The change in FEV1 in the culture based group from baseline was +0.26 L.	The change in FEV1 in the cycled group was 0L higher (0.09 L lower to 0.09 L higher).	NA	131 (1)	⊕⊝⊝⊝ very low1,2,3	The difference between the two groups was not statistically significant P = 0.97. FEV1 was also measured in percent predicted which showed a non‐significant difference between the groups at up to two years. MD 0.7% (‐4.33% to 5.73%) P = 0.79.
Pulmonary exacerbations: frequency Follow‐up: up to 583 days	The mean frequency of pulmonary exacerbation in the culture‐based group was 1.1.	The mean frequency of pulmonary exacerbation in the cycled group was 0.18 pulmonary exacerbations per person lower (0.51 lower to 0.14 higher).	NA	253 (1)	⊕⊕⊕⊝ moderate1	The difference between the groups was not statistically significant, P = 0.27.
Time to chronic PA	This outcome was not measured.
Adverse events: severe adverse events (total) Follow‐up: 18 months	231 per 1000	150 per 1000 (90 to 256)	RR 0.65 (0.39 to 1.11)	253 (1)	⊕⊕⊕⊝ moderate1	There was no significant difference between groups for the total number of participants experiencing a serious adverse event P = 0.11.
Emergence of novel bacteria: isolation of novel gram negative organisms Follow‐up: median 494 days	276 per 1000	276 per 1000 (185 to 414)	RR 1.0 (0.67 to 1.5)	253 (1)	⊕⊕⊕⊝ moderate1	No significant difference in the acquisition of novel organisms between treatment groups. P = 0.98
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CF: cystic fibrosis; CI: confidence interval; HR: hazard ratio; RR: risk ratio; PA: Pseudomonas aeruginosa.
GRADE Working Group grades of evidence High quality: further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: we are very uncertain about the estimate.

1. Downgraded once due to indirectness. The trial only includes children between the ages of one and 12 years and was not designed to answer the specific question posed in this review.

2. Downgraded once due to risk of bias from incomplete outcome data. Although this was taken from individual patient data, only 131 of the 253 participants included in our subset, were included in this analysis.

3. Downgraded once from small participant numbers which do not meet the optimum information size.

Background

Description of the condition

Cystic fibrosis (CF) is the commonest autosomal recessive, life‐limiting condition in white populations, affecting between 70,000 (Cystic Fibrosis Foundation 2015) and 100,000 people (Cystic Fibrosis Trust 2015) worldwide. It is caused by defects in the gene coding for an epithelial ion channel, the cystic fibrosis transmembrane conductance regulator (CFTR). Clinically, the disease affects multiple systems with most of the morbidity and mortality stemming from the respiratory effects. A cycle of recurrent infection, inflammation and progressive lung damage ultimately results in respiratory failure which is the primary cause of death (Cystic Fibrosis Foundation 2018). The current median age of death in people with CF in the UK is 31 years (Cystic Fibrosis Trust 2018) and 30.7 years in the USA (Cystic Fibrosis Foundation 2018), although the life expectancy of children born in 2000 is predicted to be over 50 years (Dodge 2007).

Respiratory infections in CF occur from infancy. By the age of three years 50% of children will have had an initial infection with Pseudomonas aeruginosa (PA) (Jones 2015). In the UK, half of people with CF (pwCF) have chronic infection with PA by their mid‐twenties (Cystic Fibrosis Trust 2018).

If PA is neither spontaneously cleared nor eradicated with antibiotic therapy, the CF lung environment facilitates PA adaptation to a mucoid phenotype (bacteria embedded in a sticky gel) (Döring 2010; Pitt 1986). These organisms become embedded in an exopolysaccharide biofilm which protects the organism from phagocytosis (ingestion of the bacteria by an immune cell) and reduces the efficacy of antimicrobial agents (Speert 1990). Once this change has occurred, the mucoid PA is virtually impossible to eradicate (Burns 2001; Döring 2010; Hogardt 2010; Høiby 2005; Pitt 1986). If the PA infection cannot be cleared, the affected person is faced with an increased treatment burden, accelerated decline in lung function (Emerson 2002; Konstan 2007; Kosorok 2001), increased symptom severity (Nixon 2001; Zemanick 2015) and increased mortality (Emerson 2002).

Acquisition of PA

PA is an environmental organism and an opportunistic human pathogen ‐ infection typically occurs where host defences are weakened. Early isolates usually have features common to environmental organisms (Burns 2001); therefore, the commonest source of infection is believed to be environmental. The organism is present in soil (Peeters 2016) and both fresh and salt water (Khan 2007; Pirnay 2005). Strains of PA identical to those isolated from newly infected pwCF were found in the homes of nine out of 50 pwCF; however, the temporal relationship of these isolates is unknown (Schelstraete 2008). Other potential sources identified are hot tubs (Govan 1992) and dental equipment (Barben 2008; Jensen 1997; Mainz 2015).

Environmental conditions may affect PA acquisition. Psoter found an association between increasing exposure to fine particulate matter in the atmosphere and earlier acquisition of PA (Psoter 2015). Ambient temperature may also have an effect. Research from Denmark from the early 1990s suggested PA infection peaked in the winter months (Johansen 1992). However, in 2013 Psoter showed an increased incidence of infection in the summer and autumn (Psoter 2013). Other investigators found that pwCF living in areas with higher ambient temperatures have a higher prevalence of PA, and acquire it earlier, than those living in cooler climates (Collaco 2011).

Certain strains of PA are transmissible between pwCF (Cheng 1996; Jones 2001). This understanding led to the introduction of patient segregation, though the risk of transmission is low without long‐term close contact (Speert 1987; Speert 2002). Nosocomial PA spread in hospitals is a source of major concern, though studies have failed to find a reservoir (Jones 2003; Panagea 2005).

Isolates of PA have been identified in air samples taken from rooms with PA‐positive pwCF (Jones 2003) and these can survive in airborne droplet nuclei (less than 10 μm diameter), supporting a possible airborne route of PA transmission (Clifton 2008). Knibbs subsequently demonstrated that cough aerosols generated by pwCF positive for PA contain viable organisms, which remain detectable at a distance of four metres and after 45 minutes (Knibbs 2014).

Detection of PA

Microbiological samples are commonly collected from pwCF at routine clinic appointments and at the time of a pulmonary exacerbation. The frequency of this sampling depends on national practices. Early identification of PA infection allows for eradication therapy, since at acquisition the bacterial load is usually low and the organism non‐mucoid and relatively sensitive to antibiotics (Burns 2001; Döring 2010).

The preferred sampling method is a sputum sample, but in many cases this is not possible. Where the individual does not produce sputum, options include cough swabs, oropharyngeal culture (OPC), induced sputum and bronchoalveolar lavage (BAL) (Cystic Fibrosis Trust 2009).

Non‐sputum techniques have a number of disadvantages. The gold standard is BAL, but this is invasive and requires sedation or an anaesthetic and selective sampling may miss organisms. Any OPC positive for PA may reflect oral flora, rather than being truly representative of the lower airways (Rosenfeld 1999). The diagnostic accuracy of OPC cultures is poor (Ramsey 1991), but many studies continue to use OPC sampling for practical reasons. Induced sputum has been performed in children as young as six months and samples can be collected successfully in 84% of procedures. Sensitivity is as good as two lobe bronchoalveolar lavage (Ronchetti 2018).

For the purposes of this review therefore, detection of PA is defined as the detection of PA in any respiratory sample; however, PA serology alone will not be accepted as evidence of new infection. First infection with PA is defined as the lifetime first identification of PA in any respiratory sample.

Eradication of PA

Early antibiotic therapy is effective in reducing the chance of chronic infection. A Cochrane Review showed that a number of antibiotic regimens are more effective than no treatment at eradicating PA, with an effect that can be sustained for up to two years (Langton Hewer 2017). The current review follows on from this work, focusing on additional treatments given after successful eradication to prevent or delay recurrent PA acquisition in pwCF.

Many studies of PA eradication use a negative respiratory culture at the end of the active treatment period to define successful eradication. The definition of eradication is crucial to understand whether a subsequent positive PA sample is truly a new infection, rather than an incompletely cleared index episode.

Up to 90% of individuals will clear their first infection, but further episodes of intermittent infection commonly occur over the subsequent months to years until chronic infection is finally established (Høiby 2005). This recurrence may be with a new strain, or with the same strain of PA, indicating either ineffective eradication or re‐infection from a common source. One observational study reported that 19 pwCF followed up after successful eradication had recurrent infection within a median (standard deviation (SD)) time of 8 (5.7) months (range 3 to 25 months) (Munck 2001). In 14 out of the 19 pwCF, the new PA strain had a distinct genotypic profile and researchers concluded that the initial eradication therapy had successfully eradicated the PA infection, with the subsequent infection considered a separate event (Munck 2001). In contrast, in a further study of 41 pwCF who were followed after a first ever PA isolation, 18 re‐acquired PA after a median of 7.5 months (range 2 to 55 months), 11 of whom had identical genotypes at the second isolate (Schelstraete 2010).

Description of the intervention

The effect of antibiotic regimens to eradicate initial PA infection may be sustained for up to two years, but the risk of a future episode of infection remains (Langton Hewer 2017). Currently, individuals who have successfully eradicated PA revert to their pre‐eradication treatment regimens and no ongoing secondary prevention is attempted.

Secondary prevention is the early identification and treatment of health problems prior to the appearance of symptoms (Institute for Work and Health 2015). This contrasts with primary prevention in which mechanisms are applied to the whole population or a targeted, at‐risk, population to prevent an illness or event occurring. Examples of secondary prevention in the setting of infection include cotrimoxazole prophylaxis to prevent pneumonia caused by Pneumocystis jiroveci in immunosuppressed patients (Cheung 1994) or the use of antiviral agents to prevent recurrent episodes of herpes simplex infection in people with HIV (Nelson 2011).

Host factors, such as sputum rheology and mucus plugging, may be equally important in the re‐acquisition of PA. Interventions which affect these host factors, e.g. recombinant dornase alfa (rhDNase) or hypertonic saline may also have an effect on recurrence. Treatment of eligible pwCF with CFTR modulators may have an effect on the acquisition and eradication of PA; in a study from the USA the chance of a person having positive PA cultures was significantly reduced after commencing treatment with ivacaftor (Heltshe 2015). The administration of a further course of oral, inhaled or intravenous antibiotics, after eradication of PA is complete, may reduce the risk of recurrent infection. It is possible that CFTR modulators may affect PA isolation after eradication (Heltshe 2015). Finally, immunotherapy such as IgY (derived from the eggs of hens immunised against PA) may have a role in secondary prevention (Kollberg 2003).

How the intervention might work

In people with established PA infection, antibiotics are used to reduce inflammation, to maintain lung function and to reduce the chance of a pulmonary exacerbation (Cystic Fibrosis Trust 2009; Mogayzel 2014). These antibiotics are administered orally, via nebuliser or in combination with the addition of regular cycled intravenous antibiotics as the disease course progresses (Cystic Fibrosis Trust 2009).

In a retrospective study looking at nebulised gentamicin as primary prevention for the acquisition of PA, children meeting high‐risk criteria for the development of PA infection were treated for a period of three years (Heinzel 2002). In the event of a further high‐risk incident, treatment was continued until three years after the last high‐risk event. A final audit of these children in 1999 showed that all those who continued on inhaled gentamicin remained free of PA, while seven out of 16 children who stopped prophylaxis (preventative treatment) developed chronic infection (P = 0.01).

Secondary prevention in the form of a similar prolonged course of treatment could sustain the benefits achieved by eradication, delaying the development of chronic PA infection and the subsequent negative consequences.

Why it is important to do this review

Long‐term antibiotic therapy may be beneficial in delaying chronic PA infection, but it exposes individuals to the risk of adverse events (side effects of treatment, e.g. allergy, damage to hearing or kidney function). Furthermore, the burden of treatment is one reason that PA is feared by people with CF and their families (Palser 2016). The use of prophylactic antibiotics, particularly nebulised, in people with established chronic PA can impact on school, work and social life (Conway 1996). As such, strong evidence as to the efficacy of long‐term antibiotic treatment is essential before it can be advocated to people with CF. In the current financial climate it is also important to assess the cost‐benefit of such potential treatments, particularly as the prognosis of people with CF continues to improve (Hurley 2014).

We will therefore examine current evidence to assess the safety, tolerability and cost‐effectiveness of secondary prevention strategies proposed to prevent recurrent PA infection in pwCF.

Objectives

To establish whether secondary prevention strategies, using antibiotics or other therapies, increase the chances of pwCF remaining free from PA infection following successful eradication therapy.

Methods

Criteria for considering studies for this review

Types of studies

Randomised controlled trials (RCTs), published or unpublished and in any language were eligible for inclusion. Quasi‐RCTs were eligible for inclusion if the review authors were satisfied that the groups were similar at baseline.

Types of participants

pwCF, diagnosed clinically or by genetic or sweat testing. Each participant must have had an episode of PA within the last six months which was successfully treated with an eradication regimen. They must have remained free of infection with PA between the end of eradication and start of treatment for ongoing prevention.

Types of interventions

In pwCF in whom PA was successfully eradicated, we compared a time‐limited course of therapy (for antimicrobials this could be oral, inhaled or intravenous or any combination of these) to prevent a recurrent infection with the organism to usual care, placebo or another therapeutic strategy. Time‐limited therapy included all treatment in which a specific duration was pre‐specified. Time‐limited therapy included regimens where the treatment was intermittent but continued at specified intervals for a defined duration. Long‐term suppressive therapy, given for an indefinite period, was not considered.

Types of outcome measures

Primary outcomes

1. Time to next isolation of PA (identified by any method, e.g. sputum culture (spontaneous or induced), BAL or OP culture and as defined by the trial investigators)

Secondary outcomes

1. Change in quality of life (QoL) from baseline (as measured by a validated tool (e.g. the Cystic Fibrosis Questionnaire‐Revised (CFQ‐R) (Quittner 2009), the Cystic Fibrosis Quality of Life Questionnaire (CFQoL) (Gee 2000) or any other validated tool)

2. Change (absolute and relative) from baseline for pulmonary function tests

   1. forced expiratory volume in one second (FEV₁) measured in both L and % predicted

   2. forced vital capacity (FVC) measured in both L and % predicted

3. Pulmonary exacerbations

   1. time to next exacerbation

   2. frequency of exacerbations

   3. number of days in hospital

4. Nutritional parameters ‐ change from baseline

   1. weight (kg) and weight centile or Z score

   2. height (cm) (children) and height centile or Z score

   3. body mass index (BMI) and BMI centile

5. Time to chronic PA infection (as defined by the trial investigators)

6. Adherence to treatment

   1. self‐reported measures (e.g. participant diaries)

   2. secondary count measures (e.g. pill counting, days of intravenous antibiotics)

   3. electronic data (e.g. downloaded nebuliser data)

7. Adverse effects of treatment

   1. mild (self‐limiting, not requiring treatment change, e.g. wheeze with inhaled therapy which settles

   2. moderate (requires treatment discontinuation, e.g. ototoxicity (damage to the ears causing hearing loss or balance problems))

   3. severe (e.g. hospitalisation or death)

8. Mortality

9. Isolation of resistant bacteria (with detection method, i.e. conventional culture or molecular techniques, described where possible)

   1. PA with a new resistance pattern

   2. methicillin‐resistant Staphylococcus aureus (MRSA)

   3. resistant gram negative organisms (e.g. Stenotrophomonas maltophilia, Burkholderia cepacia, Achromobacter xylosoxidans)

   4. other novel organisms

10. Cost effectiveness

Search methods for identification of studies

We searched for all relevant published and unpublished trials without restrictions on language, year or publication status.

Electronic searches

The Cochrane Cystic Fibrosis and Genetic Disorders Group's Information Specialist conducted a search of the Group's Cystic Fibrosis Trials Register for relevant trials using the following terms: (pseudomonas aeruginosa OR mixed infections) AND (eradication OR preventative OR unknown).
 
 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), weekly 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's website.

Date of most recent search: 28 October 2019.

We also searched the following trial registries:

• ISRCTN registry (www.isrctn.com; searched 21 August 2019);

• US National Institutes of Health Ongoing Trials Register Clinicaltrials.gov (www.clinicaltrials.gov; searched 21 August 2019);

• World Health Organization International Clinical Trials Registry Platform (WHO ICTRP) (apps.who.int/trialsearch; searched 21 August 2019).

For details of our search strategies, please see the appendices (Appendix 1).

Searching other resources

Two review authors (SP and SS) checked the bibliographies of all included trials and any relevant systematic reviews identified for further references to relevant trials. The review authors contacted the chief investigator of any included trials for unpublished data.

Data collection and analysis

Where the review authors were unable to use all the analysis methods described below due to insufficient trials they plan to do so in future should sufficient trials be identified.

Selection of studies

One review author (SP) screened the titles and abstracts of identified trials for inclusion in this review according to the processes set out in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011a); one of two further authors (AA or SS) then also screened these references. The authors excluded trials which were obviously irrelevant and removed duplicates. No trials identified required translation into English, but the authors planned to arrange this if required.

Two authors (SP and SS) independently screened the identified trials against the review's eligibility criteria and collated multiple reports of the same trial where necessary. They were not blinded to the trial authors. They resolved any disagreement by discussion and in consultation with a third author (AS) where necessary. They contacted the corresponding trial investigator(s) for further information where it was necessary to decide whether to include a trial in the review.

Data extraction and management

Two review authors (SS and SP) independently extracted data using a data collection form which was agreed by all review authors. The data collection form included information on the trial authors and eligibility, in addition to the trial methods (type of trial, blinding, setting, duration, number of centres and dropouts) and outcome data. For each trial the review authors documented the length of time after first isolation of PA that a participant could be randomised and also the length of time after successful eradication that a participant could be randomised. They additionally recorded the active intervention (antibiotic type, route of administration, dose, duration) and the control intervention. They considered all antibiotic regimens together; if they had identified sufficient trials, they would have undertaken a subgroup analysis of administration route. They collected data on the participant demographics, including information on participants who dropped out.

The review authors collected data from the text, tables and online supplements where appropriate and resolved any disagreements by discussion. Only a subset of participants from the included trial were eligible for inclusion in this review, so the review authors requested individual participant data (IPD) from the trial authors and one author (SP) analysed these using Microsoft Excel (MS Excel 2016) and Stata version 16 (Stata 2019) to determine the baseline characteristics of the subset of participants eligible for inclusion and their outcome data.

The authors planned to present data at two weeks, one month, three months, six months and one and two years. For the included trial we have reported the time points 'up to three months', 'up to six months', 'up to one year' and 'up to two years' and entered the data into the Review Manager software (Review Manager 2014).

Assessment of risk of bias in included studies

Two authors (SP and SS) independently assessed the included trial for any risks of bias using the Cochrane's Risk of Bias tool described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011b). This tool facilitates the identification of bias in the following domains:

• random sequence generation;

• allocation concealment;

• blinding of participants and personnel;

• blinding of outcome assessment;

• incomplete outcome data;

• selective reporting;

• other bias.

We classified each domain as having a high, low or unclear risk of bias. We resolved any disagreement by discussion and sought the opinion of a third author (AS) where necessary.

Measures of treatment effect

For dichotomous outcomes, such as mortality, development of resistant bacteria and adverse events, the review authors sought data on the number of participants with each outcome and by allocated treatment group. We planned to conduct an intention‐to‐treat analysis. Data were available to conduct an intention‐to‐treat analysis for number of participants with a pulmonary exacerbation, severe adverse events and development of resistant organisms. Adherence was calculated from returned medication so an available‐case analysis was carried out for this outcome. We calculated the risk of the outcome in treatment group compared to the risk of the outcome in the control group (pooled risk ratio (RR) and its 95% confidence interval (CI)).

For continuous data, such as change in QoL scores, pulmonary function tests and nutritional parameters, the review authors calculated the pooled mean difference (MD) and its 95% CI between the intervention and control groups. If in the future different trials use different scales, they planned to calculate the pooled standardised mean difference (SMD). The authors planned to report skewed data narratively.

The review authors analysed common count data (such as number of days in hospital) as continuous data, as described in theCochrane Handbook of Systematic Reviews of Interventions (Deeks 2011).

For time‐to‐event data (such as time to next isolation of PA) the review authors calculated the pooled hazard ratios (HRs) and their 95% CIs using the Cox Proportional Hazards model in Stata version 16 (Stata 2019). They then created forest plots using the generic inverse variance method in RevMan (Review Manager 2014).

Unit of analysis issues

The review authors have not included any cross‐over trials* as this is an inappropriate design for the review question. It is firstly unlikely that all participants would have fulfilled the primary outcome (time until a new growth of PA) at the point of cross‐over; and secondly, once the outcome was reached cross‐over would be meaningless. Cluster‐randomised trials* are also inappropriate as there may be geographical differences between the PA strains which could affect eradication and re‐acquisition rates. The review authors planned to analyse factorial trials, where there was no suggestion of an interaction between the two interventions, separately (Higgins 2011c). If we had included a trial which compared multiple treatment arms of interest we would have presented these in separate comparisons. We planned to directly compare treatments of differing durations as each represents a separate treatment regimen.

* For definitions of these types of trial please see the Cochrane online glossary (community‐archive.cochrane.org/glossary).

Dealing with missing data

The review authors contacted the trial authors for missing data and received IPD for all the trial participants who had consented to data sharing. We then extracted the data for the relevant subset of participants for analysis. We attempted to collect data on the number of participants with each outcome according to the allocated group, allowing an intention‐to‐treat analysis. Where necessary, the authors planned to use the methods described in chapter 7 of the Cochrane Handbook for Systematic Reviews of Interventions to impute these data (Higgins 2011a). If we had needed to impute significant amounts of data, we would have undertaken a sensitivity analysis to compare the effects of the imputed data against the available case data.

For the included trial the review authors analysed the primary outcome measure on an intention‐to‐treat analysis, including all the participants who met the review's inclusion criteria. This was also possible for 'time to next pulmonary exacerbation', 'adverse effects', and 'isolation of resistant pathogens'. For each time point the authors presented an available‐case analysis, including all participants for whom data were available at that time point.

Assessment of heterogeneity

Had there been sufficient trials to undertake a meta‐analysis, the review authors planned to test for heterogeneity using the I² statistic, and interpret this according to the Cochrane Handbook for Systematic Reviews of Interventions (Deeks 2011). We note that the usefulness of I² depends on the magnitude and direction of the intervention effects and the strength of the evidence of heterogeneity. We would have considered an I² value of 0% to 30% to represent little or no heterogeneity, 30% to 50% to represent moderate heterogeneity, 50% to 75% to represent substantial heterogeneity and above 75% to represent considerable heterogeneity.

Assessment of reporting biases

The review authors aimed to minimise the effects of reporting biases through a number of strategies. We assessed publication bias by a comprehensive search of grey literature and clinical trials databases, as well as by discussion with researchers in the field in an attempt to identify unpublished data; this would also have helped reduce location and citation bias. We are, however, aware of the potential bias inherent in the inclusion of unpublished data. The eligibility of inclusion of trials published in any language reduced the risk of language and location bias. The review authors carefully screened trials at inclusion to look for evidence of duplicate publication, including author names, sites, interventions and participant characteristics.

We assessed outcome reporting bias by comparing outcomes specified in the 'Methods' section to those reported in the results of the full trial paper. We further investigated this bias by comparing the outcomes reported in the full paper to those stated in the published protocol (Treggiari 2011). We further compared the stated outcomes to those published on clinicaltrials.gov. Had any further concerns remained we planned to contact the trial authors to request the original trial protocol.

Had the review authors identified a sufficient number of trials (i.e. at least 10), we planned to construct funnel plots comparing trial effect to trial size. We planned to visually inspect these plots for evidence of asymmetry and, where appropriate, test for asymmetry as discussed in the Cochrane Handbook for Systematic Reviews of Interventions (Sterne 2011). We are aware that other causes of asymmetry in a funnel plot are possible and would have considered the impact of these other factors.

Data synthesis

If the review authors had been able to combine sufficient trials, we planned to assess the extracted data using a fixed‐effect meta‐analysis. In the case of substantial heterogeneity (I² greater than 50%), we would have conducted a random‐effects meta‐analysis. We present the data from the single included trial using a fixed‐effect analysis.

Subgroup analysis and investigation of heterogeneity

The review authors planned to undertake the following subgroup analyses as appropriate:

• comparison of the effect of route of antibiotic administration (oral versus inhaled versus intravenous);

• comparison between participants who underwent eradication of a first episode of PA infection versus those with previous PA infection;

• comparison of differing methods of PA detection;

• comparison of differing definitions of recurrent PA infection;

• comparison of differing definitions of chronic PA infection.

Sensitivity analysis

If appropriate the review authors would have undertaken a sensitivity analysis to ascertain whether the results of the review are robust. We would have excluded trials assessed as having a high risk of bias (more than 50% of domains with a high risk) and would have repeated the analysis to see if this has any effect on the results. If the sensitivity analysis had shown little difference, there would have been greater confidence in the results. Furthermore, in situations where the authors had made arbitrary decisions, such as the time points for analysis, or if we had imputed significant amounts of data, we would have carried out a sensitivity analysis to assess the impact of these decisions. Once again similar results would have strengthened the conclusions of this review, while conversely a marked difference would mean the review results would need to be interpreted more cautiously.

A sensitivity analysis was carried out to ascertain whether the authors decision to define eradication in the included trial as PA culture negativity at visit 2 had any effect on the primary outcome.

Summary of findings table

The review authors constructed a summary of findings table for the single comparison in the review, cycled therapy versus culture‐based therapy. They considered the following outcomes:

1. time to next isolation of PA;

2. QoL;

3. FEV₁ (change from baseline);

4. frequency of pulmonary exacerbations;

5. time to chronic PA infection;

6. adverse events; and

7. emergence of novel bacteria.

We used the GRADE approach, described in chapter 12 of the Cochrane Handbook of Systematic Review for Interventions (Schünemann 2011) to classify the body of evidence for each outcome as high, moderate, low or very low. We downgraded the quality of the evidence across five domains; risk of bias, indirectness, inconsistency, imprecision and publication bias. Where there was serious risk of bias we downgraded by one level and where it was very serious we downgraded it by two levels. Where we judged the evidence not to be high quality, we described the rationale for this judgement in footnotes to the table.

Results

Description of studies

Please see the tables for the characteristics of the included trials (Characteristics of included studies; Characteristics of excluded studies; Characteristics of studies awaiting classification; Characteristics of ongoing studies).

Results of the search

The results of the searches are presented in a PRISMA diagram (Figure 1). We identified 371 unique references to 155 trials from electronic searches and one further trial was identified following discussion with the lead author of another trial. No further trials were identified from searches of other trial databases or the reference lists of other trials.

1.

Study flow diagram.

Following review of title and abstract, we excluded 124 trials and obtained the full texts for the remaining 31 trials. One trial was included for analysis (Treggiari 2011) and one trial is ongoing (Larsson 2011). The trial identified through discussion with the lead author is listed as 'Awaiting classification' since only a subset of participants for this trial will be eligible for inclusion; IPD have been requested, but are not available at this time (OPTIMIZE 2018). The remaining 28 trials were excluded with reasons (Characteristics of excluded studies).

Included studies

One trial has been identified for inclusion in this review (Treggiari 2011).

Trial design

The trial is a factorial design primarily looking at strategies to eradicate PA (Treggiari 2011). It was randomised 1:1:1:1 to each of four treatment regimens. Culture‐based therapy versus cycled therapy was open label, but the addition of ciprofloxacin or placebo to the tobramycin solution (TIS) regimen was blinded; outcomes assessors were blinded to all treatment allocation. This was a multicentre trial conducted at 55 CF centres in the USA. The planned duration of follow‐up was 18 months, which was achieved.

Participants

Investigators randomised 306 participants (Treggiari 2011); 304 were included in the trial intention‐to‐treat analysis as two participants were subsequently found to have failed screening. The gender split was approximately equal; 154 participants were female (150 male). The trial did not include adults or adolescents; eligible participants were aged one to 12 years (Treggiari 2011). Participants had to have had either a lifetime first isolation of PA or a new isolation, defined as at least a two‐year absence of PA having had at least one respiratory culture examined per year. Participants could have had their PA positive sample up to six months prior to enrolment and have had up to one course of anti‐pseudomonal antibiotic therapy in that time (Treggiari 2011).

The trial did not require specific evidence of eradication prior to commencement of the treatment regimen. For the purposes of this review only participants with documented eradication were eligible; participants with a positive PA sample at visit 2 (approximately three weeks after commencement of the initial course of therapy) were therefore excluded.

There were no significant differences between the groups in any of the baseline parameters for our included subset of 253 participants (124 males and 129 females). In the cycled‐therapy group 59 out of 119 (49.6%) were female and in the culture‐based therapy group 70 out of 134 (52.2%) were female. The mean age in the cycled group was 6.0 years and in the culture‐based group was 5.6 years. Baseline weight was 21.5 kg in the cycled group and 20.3 kg in the culture‐based group; baseline height was 111.0 cm in the cycled group versus 107.9 cm in the culture‐based group. Baseline FEV₁ was 1.53 L in the cycled group and 1.51 L in the culture‐based group. No data on the individual mutations were available.

Interventions

All participants underwent an initial course of inhaled TIS 300 mg twice daily for 28 days. This could be extended for a further 28 days if samples taken at the week 3 visit remained positive for PA.

The trial compared two separate interventions in a factorial design. The first comparison was cycled therapy, in which participants received TIS 300 mg twice daily every three months regardless of culture results, compared to culture‐based therapy, in which participants received TIS 300 mg twice daily only in the three‐month periods in which their respiratory samples were positive for PA. The second comparison was between the addition of oral ciprofloxacin 15 to 20 mg/kg/dose (up to 750 mg) twice daily for 14 days with every TIS cycle versus matched placebo. Participants were therefore randomised to one of four groups: cycled therapy and ciprofloxacin; or cycled therapy and placebo; or culture‐based therapy and ciprofloxacin; or culture‐based therapy and placebo (Treggiari 2011). For this review only the cycled versus culture‐based comparison was eligible.

Outcomes

The primary outcome was the time to next pulmonary exacerbation, which was defined a priori as a pulmonary exacerbation requiring either intravenous antibiotics or hospitalisation; less severe pulmonary exacerbations were a secondary endpoint. This trial also had a primary microbiological endpoint, the proportion of respiratory samples which were positive for PA at each three‐month time‐point after the first treatment cycle (Treggiari 2011). Further secondary endpoints were safety (monitored by adverse events and audiology), changes in height, weight and lung function, additional safety measures including musculoskeletal symptoms, haematological, liver and renal profiles and the emergence of resistant PA or other new pathogens (Treggiari 2011).

Excluded studies

We excluded 28 trials from the review. Nine trials assessed participants with chronic PA infection (Carswell 1987; Conway 1985; Day 1988; Dinwiddie 1982; NCT00645788; NCT01180634; Konstan 2011a; Murphy 2004; Ramsey 1999). In two trials participants were PA negative at baseline (Connett 2015; Frederiksen 2006). Four trials compared short‐term antibiotic strategies for treating acute pulmonary exacerbations (Huang 1979; Latzin 2008; Martin 1980; Schaad 1989). Six trials studied eradication with no additional treatment once eradication had been achieved (Kenny 2009; Proesmans 2013; Taccetti 2012; TORPEDO Trial; Valerius 1991; Wiesemann 1998); and four trials had no eradication step (Brett 1992; Knight 1979; Loening 1979; Singh 2013). One trial compared three weeks versus three months of eradication treatment but did not look for eradication after the first three weeks (Frederiksen 1997). One trial was cross‐over in design comparing TIS and placebo where participants first treated with TIS could go on to open‐label treatment, but this was optional (Ratjen 2018).

One further trial required very careful consideration (Ratjen 2010). This was a trial of eradication, rather than one of secondary prevention, in which all participants underwent 28 days of eradication for PA with TIS 300 mg twice daily. After this they were randomised to either no further therapy or to a further 28 days of TIS. Randomisation was based on the results of PA serology taken at baseline. Following discussion with the lead author, it was clear that respiratory cultures were not taken at day 28 to assess eradication, as participants were still on treatment, which may have suppressed PA growth. Since there was no measure of eradication between the initial 28‐day period and the subsequent period of either no treatment or an additional 28 days TIS this trial did not meet our inclusion criteria (see Types of participants).

Ongoing studies

One ongoing trial seems to meet our inclusion criteria. A phase III trial examining the ability of IgY to prolong time to chronic PA infection following successful eradication is due to be published in 2019 (Larsson 2011). This is a double‐blind RCT comparing avian‐derived IgY antibodies in a 70 mL gargle to a volume‐matched placebo in participants with previously eradicated PA. Males and females aged five years or over, who were able to gargle and were PA negative at enrolment were eligible. A total of 164 participants were recruited from multiple European CF centres. The primary outcome for this trial was the time from enrolment to the first PA positive sputum, throat cough swab or endolaryngeal suction culture. Secondary outcomes included the change from baseline in FEV₁ and BMI, number of days of illness, number of days taking antibiotics, number of pulmonary exacerbations, the change in PA serum precipitins from baseline, safety and emergence of novel pathogens.

Studies awaiting classification

The OPTIMIZE trial is a multicentre (45 sites in the USA) parallel RCT examining the effect of azithromycin in preventing pulmonary exacerbations in participants who have undergone PA eradication (OPTIMIZE 2018). Participants were randomised in a 1:1 ratio, but stratified by age group. The trial was blinded four ways (participant, care provider, investigator, outcomes assessor). The trial planned for an 18‐month follow‐up but was terminated early when it reached a pre‐specified interim monitoring boundary for efficacy. The median length of follow‐up for this trial was therefore 11.8 months. Investigators enrolled 221 participants aged six months to 18 years in this trial. Participants had to have either lifetime first isolation of PA or a new isolation, defined as at least a two‐year absence of PA having had at least one respiratory culture examined per year. The positive PA sample had to be within 40 days of the baseline visit; TIS therapy could have started up to 14 days prior to this visit. All participants underwent an initial course of inhaled TIS 300 mg twice daily for 28 days. This could be extended for a further 28 days if samples taken at the week 3 visit remained positive for PA. The active intervention was oral azithromycin suspension at 10 mg/kg up to a maximum dose of 500 mg, three times a week for the duration of the trial; participants in the control group received a volume‐matched placebo. The primary end‐point was time to next pulmonary exacerbation. This was defined a priori as a pulmonary exacerbation treated with any mode of antibiotics (oral, inhaled or intravenous). Secondary endpoints included safety (monitored by adverse events and audiology), changes in height, weight and lung function, time to recurrence of PA after the first quarter of treatment, electrocardiogram (ECG) changes, frequency of exacerbations, frequency of PA positive cultures, rates of antibiotic usage, rates of hospitalisations and patient/parent reported changes in the Chronic Respiratory Infection Symptom Score (CRISS).

Initial results of this trial have been published; however, it is likely only a subsection of participants will be eligible for our review and IPD will not be available for analysis until the trial is published in full (OPTIMIZE 2018).

Risk of bias in included studies

We judged the trial to have an unclear overall risk of bias (Treggiari 2011). There is a low risk of bias for all domains except 'other potential sources of bias' which is unclear due to the potential for additional anti‐PA therapy to be given prior to commencing the trial (Treggiari 2011). We have summarised risk of bias in Figure 2.

2.

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

Allocation

Sequence generation

We deemed the risk of bias from sequence generation to be low (Treggiari 2011). The trial described the processes of randomisation clearly, both in the main paper and in supplementary information. Investigators generated a randomisation code using a computerised random number generator.

Allocation concealment

We deemed the risk of bias due to allocation concealment to be low since an interactive voice‐response system confirmed treatment allocation via email (Treggiari 2011).

Blinding

Blinding of participants and personnel (performance bias)

The trial blinded all participants and personnel to oral therapy (ciprofloxacin or placebo). Bayer Pharmaceuticals Inc. provided both active and placebo tablets and suspension; the placebo suspension was taste‐masked. However, in this trial the allocation to cycled therapy or culture‐based therapy was open‐label for participants and care providers (although the core trial investigators were blinded to all treatment allocation for the whole trial). In an attempt to reduce the risk of bias from the lack of blinding the trial used an a priori definition of a pulmonary exacerbation and investigators reviewed all hospitalisation records in conjunction with symptom diaries from the participants to ensure that this definition was met. In view of these additional measures we deemed the risk of bias for this domain to be low (Treggiari 2011).

Blinding of outcome assessors (detection bias)

We deemed the risk of detection bias to be low as it was explicitly stated that, "The core study investigators were blinded to all treatment allocation for the entire study" (Treggiari 2011).

Incomplete outcome data

The trial report stated the investigators used an intention‐to‐treat analysis for their primary outcome measures (Treggiari 2011). In reality this was a modified intention‐to‐treat analysis (all randomised participants who received at least one dose of the trial drug) as two randomised participants failed screening prior to starting any trial medications and were therefore excluded from the analysis; they did not describe how data were imputed (Treggiari 2011). Investigators described withdrawals; these were small numbers and were balanced across the groups for both numbers and explanation (Treggiari 2011). We therefore judged this domain to have a low risk of bias.

Selective reporting

We were able to examine the published protocol on the trials registry ClinicalTrials.gov (Treggiari 2011). In addition, we analysed the published methodology. We assessed this domain as having a low risk of bias, investigators reported all the stated outcome measures in the paper, the online supplement and provided additional data in the online trial entry (Treggiari 2011).

Other potential sources of bias

The trial was supported by industry in the provision of medications (Treggiari 2011). The paper specifically states that, "The industry sponsors had no role in the analysis, interpretation and writing of the manuscript".

The trial defined new onset PA as either a first lifetime PA isolation or isolation after two years PA‐free with at least one culture per year (Treggiari 2011). This frequency of respiratory tract sampling is low; USA Infection Control Guidelines recommend sampling be done quarterly (Saiman 2014) and UK guidelines recommend sampling every two months (Cystic Fibrosis Trust 2009). It has been shown that a single sputum culture will identify only 58% of the bacterial diversity identified though more invasive sampling (Rogers 2010); PA may therefore have been unknowingly present for longer than the time limits set between positive culture and enrolment, affecting the chance of successful eradication. In addition participants were allowed to have had a positive PA culture within six months of eradication and to have had up to one course of anti‐PA therapy prior to enrolment. This additional therapy may have affected the chance of eradication and the subsequent success of the trial regimens (Treggiari 2011).

The trial did not specifically answer the question posed in this review (Treggiari 2011). We made a practical decision to define eradication as PA negative culture at visit 2 and analysed the subset of participants who fulfilled this definition. This may have had an impact on selection bias as not all participants allocated to each group were analysed for our review and the trial was not powered for this end‐point.

Overall we judged the trial to have an unclear risk of other potential bias (Treggiari 2011).

Effects of interventions

See: Table 1

Cycled versus culture‐based therapy

Data are reported below from the eligible subset of the trial population from the only included trial (Treggiari 2011). For the whole trial cohort, 304 participants were included in the intention‐to‐treat population and data for 300 were made available for analysis (consent for data sharing was not provided by the other four).

We defined successful eradication as individuals who were PA negative at visit 2 (median 21 days). Whilst trial participants who were PA positive at this visit were allowed within the trial design to have a second course of tobramycin (up to 56 days) to maximise the chance of eradication, this would have added bias to our review since not all participants would have had the same eradication regimen. We conducted a sensitivity analysis for the primary outcome to assess the impact of this decision.

At the second trial visit 41 participants were PA positive and were excluded from the analysis. A further six participants had no result or an inconclusive result at this time point and were also excluded from the analysis. Thus the eligible subset included 253 participants, 119 in the cycled arm and 134 in the culture‐based arm. Of these, 64 out of 119 (53.8%) participants in the cycled group were in the ciprofloxacin arm of the ciprofloxacin versus placebo trial and 70 out of 134 (52.2%) participants in the culture‐based group were also in the ciprofloxacin arm.

Primary outcomes

1. Time to next isolation of PA

The time to next isolation of PA was not reported in the published trial report, but could be calculated from the IPD provided. The median follow‐up for the included participants was 494 days and in the cycled group 26 out of 119 participants had recurrent PA during follow‐up while in the culture‐based group 54 out of 134 participants experienced a recurrence of PA. We were unable to measure the median time to recurrence as fewer than 50% of participants in each group had experienced a recurrence at the end of the trial. The time to recurrent PA favoured cycled over culture‐based therapy. The HR, calculated using the Cox Proportional Hazards Model, for recurrence of PA in the culture‐based group compared to the cycled group was 2.04 (95% CI 1.28 to 3.26) (Analysis 1.1). The proportional hazards assumption was checked and found to hold (P = 0.68). The GRADE rating for this evidence was moderate; it was downgraded once for indirectness (only examined children aged one to 12 years old) (Table 1). This is in contrast to the main trial report which examined rates of PA positivity (rather than time to recurrent PA) and found no difference between the cycled an culture‐based groups (Treggiari 2011).

1.1. Analysis.

Comparison 1 Cycled versus culture‐based therapy, Outcome 1 Time to next isolation of PA in participants PA negative at visit 2.

To assess the effect of the decision to use PA negativity at visit 2 to define eradication we conducted a sensitivity analysis using PA negativity at visit 3. For this analysis 244 participants were eligible, 123 in the cycled arm and 121 in the culture‐based arm. In the cycled group 25 out of 123 (20.3%) participants had recurrent PA during follow‐up, while in the culture‐based arm 44 out of 121 (36.4%) participants experienced a recurrence. Again, the median time to recurrence could not be calculated. The HR, calculated as above using the Cox Proportional Hazards Model, for recurrence of PA in the culture‐based group compared to the cycled group was 1.98 (95% CI 1.21 to 3.23) (Analysis 1.2). 19 of 123 (15.5%) participants in the cycled group had additional TIS therapy compared to 7 out of 121 (5.8%) participants in the culture‐based group. The proportional hazards assumption was checked and found to hold (P = 0.96).

1.2. Analysis.

Comparison 1 Cycled versus culture‐based therapy, Outcome 2 Time to next isolation of PA in participants PA negative at visit 3.

Secondary outcomes

1. Change in QoL from baseline

This was not reported in the included trial (Treggiari 2011).

2. Change from baseline for pulmonary function tests

The included trial looked at participants aged between one and 12 years old (Treggiari 2011). Pulmonary function testing was only performed in participants aged four years or over who were able to perform spirometry; hence at the end of trial follow‐up visit data were only available for 130 participants.

a. FEV₁

The Treggiari paper reports the absolute mean (SD) change from baseline of FEV₁ and FEV₁ % predicted at the end of trial visit (week 70) (Treggiari 2011). Using the IPD provided we were able to analyse the mean (SD) absolute and relative change from baseline for FEV₁ L and FEV₁ % predicted at 'up to three months', 'up to six months', 'up to one year', and 'up to two years'. The GRADE of this evidence was very low. It was downgraded three times for indirectness (the results only relate to children), incomplete outcome data (only 131 out of 252 participants) and small sample size (Table 1).

There were no differences identified between the cycled or culture‐based therapies for any of the measurements of FEV₁ reported at any time point: absolute change from baseline in L (Analysis 1.3), relative change from baseline in L (Analysis 1.4), absolute change from baseline in % predicted (Analysis 1.5), or relative change from baseline in % predicted (Analysis 1.6) (all very low‐quality evidence).

1.3. Analysis.

Comparison 1 Cycled versus culture‐based therapy, Outcome 3 FEV₁ (L): absolute change from baseline.

1.4. Analysis.

Comparison 1 Cycled versus culture‐based therapy, Outcome 4 FEV₁ (L): relative change in from baseline.

1.5. Analysis.

Comparison 1 Cycled versus culture‐based therapy, Outcome 5 FEV₁ % predicted: absolute change from baseline (% predicted).

1.6. Analysis.

Comparison 1 Cycled versus culture‐based therapy, Outcome 6 FEV₁ % predicted: relative change from baseline (% predicted).

b. FVC

The published Treggiari paper does not report changes in FVC, in either L or % predicted (Treggiari 2011). Using the IPD provided we were able to analyse the mean (SD) absolute and relative change from baseline for FVC (L) at 'up to three months', 'up to six months', 'up to one year', and 'up to two years'.

There was no difference in absolute change of FVC (L) from baseline (Analysis 1.7) or relative change of FVC from baseline between the cycled and culture‐based groups at any of the time points analysed (Analysis 1.8).

1.7. Analysis.

Comparison 1 Cycled versus culture‐based therapy, Outcome 7 FVC: absolute change from baseline (L).

1.8. Analysis.

Comparison 1 Cycled versus culture‐based therapy, Outcome 8 FVC: relative change from baseline (L).

3. Pulmonary exacerbations

a. time to next exacerbation

The primary outcome of the included trial was the time to next pulmonary exacerbation (in days) meeting their a priori definition and requiring either intravenous antibiotic or hospitalisation; the investigators reported no difference between cycled therapy and culture‐based for the entire cohort, HR 0.95 (95% CI 0.54 to 1.66) (Treggiari 2011).

We used the trial's secondary outcome, pulmonary exacerbation (using the trial definition) requiring any antibiotic therapy (intravenous, inhaled or oral) for our definition of a pulmonary exacerbation (Treggiari 2011). For our subset of participants data were available for 253 participants, 119 in the cycled group and 134 in the culture‐based group. Four participants experienced a pulmonary exacerbation prior to commencement of trial medication and were excluded from analysis. Data were therefore available from 249 participants. We found no difference in the time to next pulmonary exacerbation (in days) between participants in the cycled arm and those in the culture‐based arm (Analysis 1.9). The median time to next pulmonary exacerbation was 518 days in the cycled group and 495 days in the culture‐based group. The interquartile range was not calculable for either group as 75% of participants had not undergone a pulmonary exacerbation by the end of the trial.

1.9. Analysis.

Comparison 1 Cycled versus culture‐based therapy, Outcome 9 Time to next exacerbation.

b. frequency of exacerbations

The trial reported the number of participants who experienced a pulmonary exacerbation and for our participant subset data were available for 253 participants (Treggiari 2011). There was no difference in the number of participants experiencing a pulmonary exacerbation between groups (Analysis 1.10). The GRADE rating of this evidence was moderate. It was downgraded once for indirectness as the results only relate to children (Table 1).

1.10. Analysis.

Comparison 1 Cycled versus culture‐based therapy, Outcome 10 Number of participants with a pulmonary exacerbation.

There was no difference in the frequency of exacerbations per participant between the groups (Analysis 1.11).

1.11. Analysis.

Comparison 1 Cycled versus culture‐based therapy, Outcome 11 Frequency of pulmonary exacerbations (at end of trial).

The mean (SD) frequency of additional courses of antibiotics was 2.50 (2.21) in the cycled group and 3.34 (2.78) in the culture‐based group; analysis of these data clearly favoured the cycled group, MD ‐0.85 (95% CI ‐1.46 to ‐0.23) (Analysis 1.11).

c. number of days in hospital

The trial reported number of hospitalisations, but not the number of days in hospital (Treggiari 2011).

4. Nutritional parameters

a. weight (kg) and weight centile or Z score

The Treggiari paper reports the absolute mean (SD) change in weight (kg) from baseline at the end of trial visit (week 70) (Treggiari 2011) but does not report changes in weight percentile. Using the IPD provided we were able to analyse the mean (SD) absolute change from baseline for weight and weight percentile at 'up to three months', 'up to six months', 'up to one year', and 'up to two years'.

There was no difference at any of the time points analysed between the cycled and culture‐based groups in either absolute change in weight (kg) or change in weight percentile (Analysis 1.12; Analysis 1.13) .

1.12. Analysis.

Comparison 1 Cycled versus culture‐based therapy, Outcome 12 Weight (kg): absolute change from baseline.

1.13. Analysis.

Comparison 1 Cycled versus culture‐based therapy, Outcome 13 Weight percentile: absolute change from baseline.

b. height (cm) (children) and height centile or Z score

The Treggiari paper reports the absolute mean (SD) change in height (cm) from baseline at the end of trial visit (week 70) (Treggiari 2011) but does not report changes in height percentile. Using the IPD provided we were able to analyse the mean (SD) absolute change from baseline for height and change in height percentile from baseline at 'up to three months', 'up to six months', 'up to one year', and 'up to two years'.

There was no difference in absolute change in height (cm) or change in height percentile between the cycled and culture‐based groups at any of the time points analysed (Analysis 1.14; Analysis 1.15) .

1.14. Analysis.

Comparison 1 Cycled versus culture‐based therapy, Outcome 14 Height (cm): absolute change from baseline.

1.15. Analysis.

Comparison 1 Cycled versus culture‐based therapy, Outcome 15 Height percentile: absolute change from baseline.

c. BMI and BMI centile

This outcome was not reported in the included trial (Treggiari 2011).

5. Time to chronic PA infection

This outcome was not reported in the included trial (Treggiari 2011).

6. Adherence to treatment

a. self‐reported measures

This outcome was not reported in the included trial and data were not available to calculate the results (Treggiari 2011).

b. secondary count measures

In the Treggiari trial adherence was measured by counting returned medication (Treggiari 2011). In the published paper there is a comment that, "compliance with study mediations [was] 90% or greater across treatment groups," but no data are presented. Using the IPD provided we were able to analyse the number of participants returning any trial medication and the number of participants returning tobramycin at 'up to three months', 'up to six months', 'up to one year', and 'up to two years'.

There was no difference between the cycled and culture‐based groups in the percentage of participants returning any trial medication at any of the time‐points analysed (Analysis 1.16) or in the proportion of participants returning tobramycin at any of the time‐points analysed (Analysis 1.17).

1.16. Analysis.

Comparison 1 Cycled versus culture‐based therapy, Outcome 16 Adherence: number not returning trial medication.

1.17. Analysis.

Comparison 1 Cycled versus culture‐based therapy, Outcome 17 Adherence: number not returning tobramycin.

Adherence to tobramycin (inhaled) was better than adherence to overall trial medication at all time points (Analysis 1.16; Analysis 1.17). However, the forest plot examining the number of participants who returned no trial medication appears to show a trend towards better adherence in the culture‐based group over time (Analysis 1.16).

c. electronic data

This was not reported in the included trial (Treggiari 2011).

7. Adverse effects of treatment

The published trial report only reported rates of serious adverse events; they reported no difference in the rate of these between intervention groups (Treggiari 2011).

a. mild

Mild adverse events were not reported in the included trial (Treggiari 2011).

b. moderate

Moderate adverse events were not reported in the included trial (Treggiari 2011).

c. severe

A total of 49 out of 253 (19.4%) participants in our subset experienced a severe adverse event during the course of the trial; this was not significantly different between the cycled and culture‐based groups (Analysis 1.18). There was no significant difference in the rate of severe adverse events in any of the categories studied (respiratory, infection, gastro‐intestinal, metabolism, musculoskeletal, nervous system, skin or other) (Analysis 1.18). The GRADE rating for this evidence was moderate; it was downgraded once for applicability since the participants were all children (Table 1).

1.18. Analysis.

Comparison 1 Cycled versus culture‐based therapy, Outcome 18 Number of severe adverse events.

8. Mortality

This outcome was not reported in the included trial (Treggiari 2011).

9. Isolation of resistant bacteria

a. PA with a new resistance pattern or novel mucoid PA

The trial reported on the emergence of PA newly resistant to ciprofloxacin or tobramycin and novel identification of mucoid PA (Treggiari 2011). The included subset of participants were sampled at six time points after the visit 2 sample which defined inclusion in the trial. The 119 participants in the cycled arm provided a median (range) of six (zero to six) samples; in the culture‐based arm 134 participants provided a median (range) of six (zero to six) samples.

There was no significant difference in the rate of detection of PA newly resistant to either ciprofloxacin or tobramycin between the cycled and culture‐based groups (Analysis 1.19).

1.19. Analysis.

Comparison 1 Cycled versus culture‐based therapy, Outcome 19 Isolation of newly resistant PA.

Isolation of novel mucoid PA was not significantly different between the groups (Analysis 1.20).

1.20. Analysis.

Comparison 1 Cycled versus culture‐based therapy, Outcome 20 Isolation of newly mucoid PA.

b. MRSA

The included trial reported the novel identification of Staphylococcus aureus, but this was not further specified as methicillin‐sensitive Staphylococcus aureus (MSSA) versus methicillin‐resistant Staphylococcus aureus (MRSA) (Treggiari 2011).

c. resistant gram negative organisms

The Treggiari trial reported on the emergence of novel Stenotrophomonas maltophilia, Achromobacter xylosoxidans and Burkholderia cepacia complex (Treggiari 2011). The included subset of participants were sampled at six time points after the visit two sample which defined inclusion in the trial. In the cycled arm, 119 participants provided a median (range) of six (zero to six) samples and in the culture‐based arm 134 participants provided a median (range) of six (zero to six) samples. There was no significant difference between groups (Analysis 1.21). The GRADE rating for this evidence was moderate. It was downgraded once for indirectness (Table 1).

1.21. Analysis.

Comparison 1 Cycled versus culture‐based therapy, Outcome 21 Isolation of novel resistant gram negative organisms.

d. other novel organisms

The trial did not report on the emergence of any other novel pathogens (Treggiari 2011).

10. Cost effectiveness

This outcome was not reported in the included trial (Treggiari 2011).

Discussion

Only one trial with participants who fulfilled our inclusion criteria was identified and only a subset of the 306 participants enrolled in this trial were eligible (Treggiari 2011). IPD was obtained for 300 participants, of whom 253 were eligible to be included in our review.

Summary of main results

Our analysis showed that additional treatment with inhaled tobramycin every three months (cycled therapy) following eradication of PA doubled the time to recurrence of PA in a subset of participants from the EPIC trial compared to culture‐based therapy (Treggiari 2011) (Analysis 1.1). This was not affected by changing the definition of eradication (Analysis 1.2). We deemed the quality of the evidence for our primary outcome to be moderate suggesting there is probably an improvement in time to recurrence after cycled therapy. Development of newly‐resistant PA and isolation of newly‐resistant organisms were slightly higher in the culture‐based group compared to the cycled group, but this result was not statistically significant. There was no evidence of a difference between the treatment groups in terms of pulmonary function, time to next pulmonary exacerbation, severe adverse events, nutritional parameters, adherence, isolation of newly mucoid PA or novel resistant gram negative organisms. The quality of evidence for secondary outcomes was moderate according to GRADE criteria with the exception of pulmonary function which was deemed to be very low, meaning that we are uncertain whether there is an effect. The trial did not report on QoL, time to chronic PA infection, mortality or cost‐effectiveness (Treggiari 2011).

In the published trial report, the authors concluded that there was no difference between cycled and culture‐based therapy in the rate of pulmonary exacerbation or PA positivity (Treggiari 2011). This is in contrast to the findings of our review, where we found that the time to next isolation of PA was doubled by the provision of cycled TIS therapy. The difference in these findings could be due to the fact that only a subset of the trial's enrolled participants were eligible for inclusion in our review; although the baseline characteristics of our subset were similar to those of the whole cohort described in the main trial report except that the percentage of participants positive for PA at baseline in the cycled group was only 27.4% in our subset, compared to 38% in the full cohort (Characteristics of included studies). Another explanation could be the different question posed by this review compared to the main trial, namely the time to next isolation of PA rather than the rate of PA positivity in each group (Treggiari 2011). An additional consideration is that the results in the main trial report were age‐adjusted, which is not the case for the results presented here.

Overall completeness and applicability of evidence

The included trial only included children up to 12 years of age (Treggiari 2011). Eradication can be successful in adults with CF (Langton Hewer 2017); the age at onset of chronic PA infection is increasing (Cystic Fibrosis Trust 2018) and the percentage of individuals with a PA positive culture is decreasing (Cystic Fibrosis Foundation 2018). Strategies to prevent the recurrence of PA after successful eradication are therefore important for all age groups, but the findings of this review cannot be directly applied to adolescents and adults with CF. The paucity of trials also meant that we were not able to comment on some of our outcomes (changes in QoL, time to chronic PA infection, mortality and cost‐effectiveness).

The trial allowed inclusion of participants whose PA‐positive qualifying sample occurred up to six months prior to the baseline visit. Participants could have undergone up to one course of eradication therapy prior to enrolment. This may have had an effect on the efficacy of the intervention, either biasing it towards participants who had previously undergone eradication or conversely biasing it towards participants with only a short lag between PA identification and the commencement of treatment. Results from the EARLY trial suggest that a delay in the commencement of eradication therapy for PA negatively affects the chance of success (Ratjen 2018).

The included trial followed up participants for a median of 494 days. Up to this point there was no evidence of difference between the two groups (cycled therapy or culture‐based therapy) in terms of clinical parameters (lung function, nutritional parameters, time to next pulmonary exacerbation), thus although PA recurrence was delayed in the cycled group, this was not translated into objective clinical benefit. Chronic PA infection is associated with increased morbidity and mortality, but it may be that longer follow‐up is required to truly appreciate whether cycled therapy has a clinically apparent benefit.

It is also important to note that the included trial does not directly address the question posed in this review (Treggiari 2011). The same problem applies to all currently‐designed eradication studies. Treggiari aimed to compare prolonged eradication regimens to reduce the chance of pulmonary exacerbation after the acquisition of PA. Evidence of eradication was not a requirement in the trial protocol prior to the commencement of the trial. We therefore had to use a proxy measure of eradication and could only include a subset of participants from the trial, which reduces the applicability of the evidence to this review question. The use of a single negative culture result (which may have been an OPC culture) further reduces the applicability of the evidence for two reasons. Firstly, the visit 2 sample was taken while the participant was still on treatment so a negative result may represent suppression rather than true eradication; and secondly, since a single sputum sample may only identify 58% of the bacterial diversity compared to more intensive sampling, this may be insufficient to be certain that eradication has occurred (Rogers 2010).

Quality of the evidence

We have identified only one trial to include in this review and examined data from 252 participants (Treggiari 2011), although the results of two further trials are awaited (Larsson 2011; OPTIMIZE 2018). The included trial was adequately powered to identify differences in its primary outcomes (time to first severe pulmonary exacerbation and proportion of PA positive cultures at each quarter). However, by excluding participants who did not fulfil the inclusion criteria for this review, these results may now be underpowered to show a difference. The risk of bias was deemed to be low across all domains with the exception of 'other bias'. Participants could be enrolled up to six months after a PA positive culture and could have had up to one course of eradication therapy during that time. This may have had an effect on the efficacy of the intervention, either biasing it towards participants who had previously undergone eradication or conversely biasing it towards participants with only a short lag between PA identification and the commencement of treatment. We have therefore stated that there is an unclear risk of 'other bias'.

The nature of the trial as a four‐way factorial trial meant that participants in each of the cycled and culture‐based therapy groups did not receive equal treatment. In our population subset 53.8% of participants in the cycled group and 52.2% of participants in the culture‐based group received oral ciprofloxacin in addition to their inhaled tobramycin. Further breaking down the analysis into four groups would further reduce the numbers available for each comparison.

Since this trial was conducted in young children many were unable to expectorate sputum. Most respiratory sampling in the trial relied on oropharyngeal swab cultures and the reliability of oropharyngeal cultures to reflect the microbial populations of the lower respiratory tract is limited (Armstrong 1996). This may have caused false‐positive and false‐negative results in the sampling, resulting in detection bias, though it is likely that this bias would be equally applied to both groups. Ongoing treatment with anti‐pseudomonal antibiotics theoretically may have a suppressive effect on the growth of PA such that it could not be detected, particularly in OPC. However, the interval of treatment makes this less likely, since the sputum density of PA increased almost back to pre‐treatment levels during off‐treatment intervals in a study of intermittent tobramycin administration in participants chronically infected with PA (Ramsey 1999).

Using GRADE we found that the overall quality of the evidence varied from very low to moderate across the outcomes presented in the summary of findings table. We deemed most outcomes to provide moderate‐quality evidence, downgraded only because the trial only includes children. We cannot be certain if the results would be reproducible in an adult population. We deemed the quality of evidence for pulmonary function to be very low due to the fact that younger children were unable to perform spirometry and so there was imprecision in the results from low participant numbers and missing data. Specifically, we deemed the quality of the evidence for our primary outcome to be moderate suggesting there is probably an improvement in time to recurrence after cycled therapy. The quality of evidence for secondary outcomes was moderate, with the exception of pulmonary function which was deemed to be very low meaning that we are uncertain whether there is an effect.

Potential biases in the review process

A potential source of bias in the review process is the lack of a formal definition of eradication in CF clinical trials. The included trial did not formally assess the success of initial eradication therapy; however, a sample was obtained three weeks after therapy commenced. Participants with a positive result at this stage were given an extra 28 days of TIS. We have therefore defined eradication as a negative PA sample at this three‐week time point since a later cut‐off would add another treatment variation for participants who underwent 56 rather than 28 days of TIS. However, we note that since this sample was taken whilst treatment was ongoing, suppression of PA (rather than true eradication) may have occurred. Indeed, in discussion with the authors of the ELITE trial, which we had anticipated including, it became evident that for this reason they did not send microbiological samples after the first 28 days of treatment.

Agreements and disagreements with other studies or reviews

We are not aware of any other studies or reviews which recommend strategies to prevent recurrence of PA following successful eradication.

However, it is of note that a previous Cochrane Review examining antibiotic strategies for eradicating PA in people with CF, also found that cycled therapy was superior to culture‐based therapy (Langton Hewer 2017). The authors of that review used different primary outcomes to ours and included adjustment for participant age in their analyses, which may account for some of the differences found.

Authors' conclusions

Implications for practice.

From the limited evidence available, cycled therapy appears to delay the recurrence of Pseudomonas aeruginosa (PA) after eradication in children. Further study is, however, required to better understand the benefits of cycled therapy in adolescents and adults with cystic fibrosis (CF).

The burden of additional nebulised therapy needs to be weighed against its efficacy. We were unable to examine the effect of cycled therapy compared to culture‐based therapy on treatment burden as quality of life was not measured. Adding 28 days of nebulised therapy every three months for 15 months may seem a small addition to a therapeutic regimen, but we should bear in mind that the addition of nebulised therapy is often cited by people with CF as having a large impact on treatment burden (Palser 2016).

Implications for research.

Now that eradication of PA is commonplace in the management of CF, further consideration needs to be made in terms of the next steps. We suggest that randomised controlled trials, specifically dedicated to investigating the prevention of PA recurrence, are needed. These trials must be conducted in participants of all ages with CF. Treatments other than antimicrobial agents, which reduce the chance of recurrence by improving host defence or exposure to the organism, could also be considered.

Success in treating infection in CF may necessitate the use of surrogate markers, as evidenced by Treggiari (Treggiari 2011). In this trial, the median time to recurrence of PA could not be determined in a trial with a planned 15‐month follow‐up and time to chronic PA would require even longer follow‐up. Time to next pulmonary exacerbation may be the most pragmatic outcome and would facilitate comparisons with other CF pathogens. However, as discussed above, the use of time to recurrence as the primary outcome measure meant that Treggiari reported no difference in the two treatment groups; whilst we have shown a hazard ratio of 2.04 in the time to recurrence of PA for those undergoing culture‐based therapy compared to cycled therapy.

In addition to time to chronic PA infection and time to next pulmonary exacerbation, outcomes to be included in future trials of interventions to prevent PA recurrence should include, lung function, nutritional parameters, cost‐effectiveness and quality of life including treatment burden.

Since delaying the onset of chronic PA infection may reduce the longer‐term morbidity and mortality of CF, assessing these outcomes would require longer durations of observation than are feasible in a traditional RCT. We therefore suggest consideration of alternative trial designs such as registry‐based trials may be the best way to examine these outcomes.

These trials need to use internationally standardised definitions for:

1. chronic PA infection;

2. pulmonary exacerbations;

3. eradication of PA.

Development and widespread use of these definitions will improve the comparability of trials, strengthening the conclusions which can be drawn when they are compared and would form part of a core outcome set for CF. The importance of a core outcome set to reduce selective reporting bias and thereby improve the validity of systematic review in CF has previously been highlighted (Dwan 2013). The results of the Core Outcome Set Taskforce for CF (COST‐CF) will be an important development in the care of people with CF (Smyth 2017).

Appendices

Appendix 1. Search methods ‐ electronic searches

Database/Resource	Strategy
ISRCTN registry	[Advanced Search] TEXT SEARCH: pseudomonas OR aeruginosa OR infection CONDITION: cystic fibrosis
Clinicaltrials.gov	[Advanced Search] Search 1: CONDITION/ DISEASE: pseudomonas OR aeruginosa OTHER TERMS: eradicate OR eradication OR eradicating OR prevent OR prevention OR preventing OR preventative OR reoccurrence OR recur OR recurrent OR recurrence OR reoccur STUDY TYPE: Interventional Studies Search 2: CONDITION/ DISEASE: cystic fibrosis AND infection OTHER TERMS: eradicate OR eradication OR eradicating OR prevent OR prevention OR preventing OR preventative OR reoccurrence OR recur OR recurrent OR recurrence OR reoccur STUDY TYPE: Interventional Studies
WHO ICTRP	[Advanced Search] TITLE: eradicate OR eradication OR eradicating OR prevent OR prevention OR preventing OR preventative OR reoccurrence OR recur OR recurrent OR recurrence OR reoccur AND CONDITION: cystic fibrosis RECRUITMENT STATUS: all

Data and analyses

Comparison 1. Cycled versus culture‐based therapy.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Time to next isolation of PA in participants PA negative at visit 2	1		Hazard Ratio (Fixed, 95% CI)	Subtotals only
1.1 At up to 2 years	1	253	Hazard Ratio (Fixed, 95% CI)	2.04 [1.28, 3.26]
2 Time to next isolation of PA in participants PA negative at visit 3	1		Hazard Ratio (Fixed, 95% CI)	Subtotals only
2.1 At up to 2 years	1	244	Hazard Ratio (Fixed, 95% CI)	1.98 [1.21, 3.23]
3 FEV1 (L): absolute change from baseline	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
3.1 Up to 3 months	1	132	Mean Difference (IV, Fixed, 95% CI)	‐0.04 [‐0.11, 0.04]
3.2 Up to 6 months	1	130	Mean Difference (IV, Fixed, 95% CI)	0.02 [‐0.05, 0.10]
3.3 Up to 1 year	1	131	Mean Difference (IV, Fixed, 95% CI)	‐0.02 [‐0.10, 0.06]
3.4 Up to 2 years	1	131	Mean Difference (IV, Fixed, 95% CI)	‐0.00 [‐0.09, 0.09]
4 FEV1 (L): relative change in from baseline	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
4.1 Up to 3 months	1	132	Mean Difference (IV, Fixed, 95% CI)	‐0.02 [‐0.09, 0.05]
4.2 Up to 6 months	1	130	Mean Difference (IV, Fixed, 95% CI)	0.02 [‐0.05, 0.10]
4.3 Up to 1 year	1	131	Mean Difference (IV, Fixed, 95% CI)	0.01 [‐0.08, 0.11]
4.4 Up to 2 years	1	131	Mean Difference (IV, Fixed, 95% CI)	0.02 [‐0.08, 0.12]
5 FEV1 % predicted: absolute change from baseline (% predicted)	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
5.1 Up to 3 months	1	132	Mean Difference (IV, Fixed, 95% CI)	‐2.41 [‐6.90, 2.08]
5.2 Up to 6 months	1	130	Mean Difference (IV, Fixed, 95% CI)	2.06 [‐2.66, 6.77]
5.3 Up to 1 year	1	130	Mean Difference (IV, Fixed, 95% CI)	‐0.37 [‐5.07, 4.32]
5.4 Up to 2 years	1	130	Mean Difference (IV, Fixed, 95% CI)	0.70 [‐4.33, 5.73]
6 FEV1 % predicted: relative change from baseline (% predicted)	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
6.1 Up to 3 months	1	132	Mean Difference (IV, Fixed, 95% CI)	‐0.02 [‐0.09, 0.05]
6.2 Up to 6 months	1	130	Mean Difference (IV, Fixed, 95% CI)	0.03 [‐0.05, 0.10]
6.3 Up to 1 year	1	130	Mean Difference (IV, Fixed, 95% CI)	0.01 [‐0.07, 0.10]
6.4 Up to 2 years	1	130	Mean Difference (IV, Fixed, 95% CI)	0.02 [‐0.07, 0.10]
7 FVC: absolute change from baseline (L)	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
7.1 Up to 3 months	1	132	Mean Difference (IV, Fixed, 95% CI)	0.00 [‐0.07, 0.08]
7.2 Up to 6 months	1	130	Mean Difference (IV, Fixed, 95% CI)	0.04 [‐0.03, 0.12]
7.3 Up to 1 year	1	131	Mean Difference (IV, Fixed, 95% CI)	‐0.01 [‐0.10, 0.08]
7.4 Up to 2 years	1	131	Mean Difference (IV, Fixed, 95% CI)	0.04 [‐0.06, 0.15]
8 FVC: relative change from baseline (L)	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
8.1 Up to 3 months	1	132	Mean Difference (IV, Fixed, 95% CI)	0.01 [‐0.06, 0.08]
8.2 Up to 6 months	1	130	Mean Difference (IV, Fixed, 95% CI)	0.05 [‐0.03, 0.12]
8.3 Up to 1 year	1	131	Mean Difference (IV, Fixed, 95% CI)	0.02 [‐0.08, 0.12]
8.4 Up to 2 years	1	131	Mean Difference (IV, Fixed, 95% CI)	0.05 [‐0.05, 0.15]
9 Time to next exacerbation	1		Hazard Ratio (Fixed, 95% CI)	Subtotals only
9.1 Up to 2 years	1	249	Hazard Ratio (Fixed, 95% CI)	1.13 [0.78, 1.62]
10 Number of participants with a pulmonary exacerbation	1		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
10.1 Up to 2 years	1	253	Risk Ratio (M‐H, Fixed, 95% CI)	0.91 [0.70, 1.18]
11 Frequency of pulmonary exacerbations (at end of trial)	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
11.1 Study‐defined pulmonary exacerbation	1	253	Mean Difference (IV, Fixed, 95% CI)	‐0.18 [‐0.51, 0.14]
11.2 Any additional antibiotics	1	253	Mean Difference (IV, Fixed, 95% CI)	‐0.85 [‐1.46, ‐0.23]
12 Weight (kg): absolute change from baseline	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
12.1 Up to 3 months	1	243	Mean Difference (IV, Fixed, 95% CI)	0.08 [‐0.13, 0.28]
12.2 Up to 6 months	1	240	Mean Difference (IV, Fixed, 95% CI)	0.08 [‐0.24, 0.40]
12.3 Up to 1 year	1	230	Mean Difference (IV, Fixed, 95% CI)	0.03 [‐0.46, 0.51]
12.4 Up to 2 years	1	241	Mean Difference (IV, Fixed, 95% CI)	0.29 [‐0.38, 0.97]
13 Weight percentile: absolute change from baseline	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
13.1 Up to 3 months	1	243	Mean Difference (IV, Fixed, 95% CI)	1.11 [‐1.20, 3.43]
13.2 Up to 6 months	1	240	Mean Difference (IV, Fixed, 95% CI)	‐0.60 [‐3.23, 2.03]
13.3 Up to 1 year	1	230	Mean Difference (IV, Fixed, 95% CI)	‐1.09 [‐4.51, 2.33]
13.4 Up to 2 years	1	241	Mean Difference (IV, Fixed, 95% CI)	‐0.56 [‐4.48, 3.36]
14 Height (cm): absolute change from baseline	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
14.1 Up to 3 months	1	242	Mean Difference (IV, Fixed, 95% CI)	0.01 [‐0.29, 0.31]
14.2 Up to 6 months	1	238	Mean Difference (IV, Fixed, 95% CI)	‐0.33 [‐0.74, 0.09]
14.3 Up to 1 year	1	228	Mean Difference (IV, Fixed, 95% CI)	‐0.18 [‐0.77, 0.41]
14.4 Up to 2 years	1	239	Mean Difference (IV, Fixed, 95% CI)	‐0.30 [‐1.09, 0.49]
15 Height percentile: absolute change from baseline	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
15.1 Up to 3 months	1	242	Mean Difference (IV, Fixed, 95% CI)	1.94 [‐0.35, 4.22]
15.2 Up to 6 months	1	238	Mean Difference (IV, Fixed, 95% CI)	0.18 [‐2.23, 2.59]
15.3 Up to 1 year	1	228	Mean Difference (IV, Fixed, 95% CI)	1.61 [‐1.32, 4.55]
15.4 Up to 2 years	1	239	Mean Difference (IV, Fixed, 95% CI)	2.07 [‐1.12, 5.25]
16 Adherence: number not returning trial medication	1		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
16.1 Up to 3 months	1	243	Risk Ratio (M‐H, Fixed, 95% CI)	1.20 [0.84, 1.71]
16.2 Up to 6 months	1	120	Risk Ratio (M‐H, Fixed, 95% CI)	1.74 [0.63, 4.82]
16.3 Up to 1 year	1	118	Risk Ratio (M‐H, Fixed, 95% CI)	1.02 [0.53, 1.98]
16.4 Up to 2 years	1	115	Risk Ratio (M‐H, Fixed, 95% CI)	0.78 [0.45, 1.36]
17 Adherence: number not returning tobramycin	1		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
17.1 Up to 3 months	1	250	Risk Ratio (M‐H, Fixed, 95% CI)	0.94 [0.76, 1.16]
17.2 Up to 6 months	1	120	Risk Ratio (M‐H, Fixed, 95% CI)	1.39 [0.83, 2.32]
17.3 Up to 1 year	1	119	Risk Ratio (M‐H, Fixed, 95% CI)	0.80 [0.63, 1.01]
17.4 Up to 2 years	1	122	Risk Ratio (M‐H, Fixed, 95% CI)	0.87 [0.71, 1.08]
18 Number of severe adverse events	1		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
18.1 Respiratory	1	253	Risk Ratio (M‐H, Fixed, 95% CI)	0.75 [0.32, 1.77]
18.2 Infections	1	253	Risk Ratio (M‐H, Fixed, 95% CI)	1.13 [0.23, 5.47]
18.3 Gastro‐intestinal	1	253	Risk Ratio (M‐H, Fixed, 95% CI)	0.42 [0.11, 1.56]
18.4 Metabolism	1	253	Risk Ratio (M‐H, Fixed, 95% CI)	0.38 [0.02, 9.12]
18.5 Musculoskeletal	1	253	Risk Ratio (M‐H, Fixed, 95% CI)	0.23 [0.01, 4.64]
18.6 Nervous system	1	253	Risk Ratio (M‐H, Fixed, 95% CI)	0.38 [0.04, 3.56]
18.7 Skin	1	253	Risk Ratio (M‐H, Fixed, 95% CI)	0.56 [0.05, 6.13]
18.8 Other	1	253	Risk Ratio (M‐H, Fixed, 95% CI)	0.73 [0.36, 1.49]
18.9 Total	1	253	Risk Ratio (M‐H, Fixed, 95% CI)	0.65 [0.39, 1.11]
19 Isolation of newly resistant PA	1		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
19.1 Up to 2 years	1	253	Risk Ratio (M‐H, Fixed, 95% CI)	0.19 [0.02, 1.54]
20 Isolation of newly mucoid PA	1		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
20.1 Up to 2 years	1	253	Risk Ratio (M‐H, Fixed, 95% CI)	0.38 [0.08, 1.82]
21 Isolation of novel resistant gram negative organisms	1		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
21.1 Up to 2 years	1	253	Risk Ratio (M‐H, Fixed, 95% CI)	1.00 [0.67, 1.50]

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Treggiari 2011.

Methods	RCT. Four‐way factorial design. Blinding for ciprofloxacin vs placebo but not for cycle‐based vs culture‐based therapy. Duration: 18 months (with subsequent optional enrolment into an ongoing observational study). Multicentre: 55 CF centres in the USA.
Participants	Eligibility criteria Diagnosis of CF and new (defined as lifetime first or at least 2 year absence of PA on culture) isolation of PA within last 6 months. Males and females aged ≧ 1 year and ≦ 12 years who were clinically stable at time of randomisation. No more than one course of intravenous or inhaled anti‐PA antibiotics in the 2 years prior to baseline. Participant characteristics ‐ published study population 306 participants randomised; 2 screen failures so 304 randomised and treated with 76 into each of the 4 treatment arms. Cycled therapy vs culture‐based therapy Age, mean (SD) years: cycled 5.8 (3.4); culture‐based: 5.5 (3.7). Gender split, number (%): cycled 77 (51) females; culture‐based 77 (51) females. Weight, mean (SD) kg: cycled 21.2 (9.9); culture‐based: 20.1 (10.3). Mutation, F508del homozygous, number (%): cycled: 66 (43); culture‐based: 83 (55). PA positive at baseline, number (%): cycled 56 (38); culture‐based 62 (42). Ciprofloxacin vs placebo Age, mean (SD) years: ciprofloxacin 5.5 (3.5); placebo: 5.9 (3.6). Gender split, number (%): ciprofloxacin 69 (45) females; placebo 85 (56) females. Weight, mean (SD) kg: ciprofloxacin 20.3 (10.3); placebo: 20.9 (9.8). Mutation, F508del homozygous, number (%): ciprofloxacin 68 (45); placebo: 81 (53). PA positive at baseline, number (%): ciprofloxacin: 58 (39); placebo: 60 (41). Please note: these characteristics refer to all the participants enrolled in the trial, rather than those analysed from individual participant data. Participant characteristics ‐ included subset of participants Data provided for 300 participants, 253 of whom fulfilled review inclusion criteria. Cycled therapy vs culture‐based therapy Age, mean (SD) years: cycled 5.5 (3.3); culture‐based: 5.1 (3.7). Gender split, number (%): cycled 59 (49.6) females; culture‐based 70 (52.2) females. Weight, mean (SD) kg: cycled 21.5 (9.9); culture‐based: 20.3 (10.5). Mutation, F508del homozygous, number (%): data not provided. PA positive at baseline, number (%): cycled 31 (27.4); culture‐based 52 (39.4).
Interventions	All participants received an initial cycle of 28 days of TIS 300 mg 2x daily + 14 days of ciprofloxacin 15 ‐ 20 mg/kg/dose up to 750 mg 2x daily or oral placebo. A further 28 days of TIS 300 mg 2x daily monotherapy could be administered if respiratory cultures at week 3 were still positive for PA. Intervention 1 (cycle‐based therapy/ciprofloxacin): scheduled TIS 300 mg 2x daily for 28 days and oral ciprofloxacin 15 ‐ 20 mg/kg/dose up to 750 mg 2x daily for 14 days every 3 months. Intervention 2 (cycle‐based therapy/placebo): scheduled TIS 300 mg 2x daily for 28 days and oral placebo for 14 days every 3 months. Intervention 3 (culture‐based therapy/ciprofloxacin): TIS 300 mg 2x daily for 28 days and oral ciprofloxacin 15 ‐ 20 mg/kg/dose up to 750 mg 2x daily for 14 days only when quarterly respiratory cultures positive for PA. Intervention 4 (culture‐based therapy/placebo): TIS 300 mg 2x daily for 28 days and oral placebo for 14 days only when quarterly respiratory cultures positive for PA.
Outcomes	Primary clinical outcome Time to first pulmonary exacerbation (using an a priori definition) requiring intravenous antibiotics or hospital admission. Primary microbiological outcome Proportion of PA positive respiratory cultures at each quarter after the 1st treatment cycle. Secondary outcomes Time to pulmonary exacerbation using any mode of antibiotics or hospitalisation. Anthropometric measures (linear growth, weight gain). Pulmonary function tests (FVC, FEF and FEV1). Safety (including adverse events, musculoskeletal symptoms, hearing acuity, haematological profile and renal and liver function). Emergence of antibiotic‐resistant PA or other significant pathogens.
Funding source	This research was partially supported by Cystic Fibrosis Foundation grant EPIC0K0, National Heart, Lung and Blood Institute and National Institute of Diabetes, Digestive and Kidney Diseases grant U01‐HL080310 and National Centre for Research Resource Grants ULI‐RR025014‐03, 1UL1‐RR025744, 1UL1RR025780, UL1‐RR025005, UL1‐RR0024979, UL1RR025747, UL1‐RR025011, 1UL‐RR024975 and M01‐RR02172. Study drugs and devices were supplied free of charge by Novartis Pharmaceutical Group (TIS), Bayer Healthcare AG (oral ciprofloxacin and oral placebo) and PARI Respiratory Equipment Inc. (nebulisers and compressors).
Declaration of interest among the primary researchers	Dr Chatfield declares research grants or contracts from Gilead Pharmaceuticals, Vertex Pharmaceuticals and the Cystic Fibrosis Foundation.
Notes	Participants could have undergone up to 1 course of anti‐pseudomonal antibiotics between PA isolation and enrolment.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Randomisation was carried out by permuted blocks and performed using a computer‐generated sequence."
Allocation concealment (selection bias)	Low risk	"Randomisation assignment was available at the sites via an interactive voice response system with email confirmation of the treatment assignment."
Blinding of participants and personnel (performance bias) All outcomes	Low risk	"All study personnel and participants were blinded to oral therapy assignment but not to cycled or culture‐based treatment allocation." Oral placebo pills or suspension were provided by the same company who provided ciprofloxacin tablets and suspension. The placebo suspension was taste‐masked. An a priori definition of pulmonary exacerbation was used and all hospitalisation records were reviewed against this definition with participant symptom diaries used as corroborating evidence.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	"The core study investigators were blinded to all treatment allocation for the entire study."
Incomplete outcome data (attrition bias) All outcomes	Low risk	The trial reports a modified intention‐to‐treat analysis was carried out to include all participants who were randomised and received at least 1 dose of trial medication. It is unclear how data were imputed. There were 27 dropouts over the trial, 8 in the cycle‐based/ciprofloxacin group, 8 in the cycle‐based/placebo group, 8 in the culture‐based/ciprofloxacin group and 3 in the culture‐based/placebo group. These were described as lost‐to‐follow‐up (n = 5), physician decision (n = 8), participant decision (n = 9), adverse event (n = 1) and other (n = 3). The number of participants with data available at each time point is stated.
Selective reporting (reporting bias)	Low risk	We have reviewed the published paper detailing the trial design and rationale which was published towards the end of data collection for this trial. We have also reviewed the ClinicalTrials.gov entry. In the methods section of the main paper time to pulmonary exacerbation requiring any antibiotic modality or hospitalisation is given as a secondary endpoint; the ClinicalTrials.gov entry only states number of participants with a pulmonary exacerbation requiring oral, inhaled or oral antibiotics. The paper reports this latter endpoint. The authors report there is no statistical difference between the groups in renal function, liver function, blood cell counts, musculoskeletal examination or hearing testing but no specific data are given in the paper. Some data but no statistical testing are given in the results published on ClinicalTrials.gov.
Other bias	Unclear risk	Participants could be enrolled up to six months after a PA‐positive culture and could have had up to 1 course of eradication therapy during that time. This may have had an effect on the efficacy of the intervention, either biasing it towards participants who had previously undergone eradication or conversely biasing it towards participants with only a short lag between PA identification and the commencement of treatment. The trial was sponsored by Novartis Pharmaceutical Corps (TIS) and Bayer Healthcare (ciprofloxacin and placebo) who provided trial drugs free of charge. Nebulisers and compressors were provided free of charge by PARI Respiratory Equipment Inc. "The industry sponsors had no role in the analysis, interpretation and writing of the manuscript." The trial did not specifically answer the question posed in this review and based on our definition of eradication as PA negative culture at visit 2, we only analysed the subset of participants who fulfilled this definition. This may have had an impact on selection bias as not all participants allocated to each group were analysed for our review and the trial was not powered for this end point.

ECG: electrocardiogram
 FEF: forced expiratory flow
 FEV₁: forced expiratory volume in 1 second
 FVC: forced vital capacity
 PA: Pseudomonas aeruginosa
 RCT: randomised controlled trial
 SD: standard deviation
 TIS: tobramycin inhalation solution
 vs: versus

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Brett 1992	Ineligible design ‐ comparison of regular vs. symptomatic treatment in participants with intermittent PA, no eradication step.
Carswell 1987	Participants had chronic PA infection.
Connett 2015	Participants were PA negative at baseline.
Conway 1985	Participants had chronic PA infection.
Day 1988	Participants had chronic PA infection.
Dinwiddie 1982	Participants had chronic PA infection.
Frederiksen 1997	Ineligible design ‐ no conclusive evidence of eradication prior to 2nd phase of treatment.
Frederiksen 2006	Participants were PA negative at baseline.
Huang 1979	Ineligible design ‐ compared antibiotic regimens for treatment of acute pulmonary exacerbation.
Kenny 2009	Ineligible design ‐ retrospective review of eradication strategies.
Knight 1979	Ineligible design ‐ cross‐over trial of cephalexin and placebo, participants probably had chronic PA, no eradication step.
Konstan 2011a	Participants had chronic PA infection.
Latzin 2008	Wrong study design ‐ comparison of IV regimens for pulmonary exacerbations.
Loening 1979	Ineligible design ‐ ongoing antibiotic treatment with no period of eradication.
Martin 1980	Ineligible design ‐ comparing short‐term IV antibiotic regimens in the management of deteriorating patients; participants probably had chronic PA.
Murphy 2004	Participants had chronic PA infection.
NCT00645788	Participants had chronic PA infection.
NCT01180634	Participants had chronic PA infection.
Proesmans 2013	Ineligible design ‐ comparison of two antibiotic regimens for eradication of PA.
Ramsey 1999	Participants had chronic PA infection.
Ratjen 2010	Ineligible design ‐ no conclusive evidence of eradication prior to 2nd phase of treatment.
Ratjen 2018	Ineligible design ‐ cross‐over trial; participants in initial active arm had no additional treatment unless they failed to eradicate PA.
Schaad 1989	Ineligible design ‐ comparing treatment regimens for acute pulmonary exacerbation; participants also had chronic PA.
Singh 2013	Ineligible design ‐ comparing symptom or culture‐based treatment of infections on rate of acquisition of PA. No eradication step.
Taccetti 2012	Ineligible design ‐ comparison of two antibiotic regimens for eradication of PA.
TORPEDO Trial	Ineligible design ‐ comparison of two antibiotic regimens for eradication of PA.
Valerius 1991	Ineligible design ‐ comparing antibiotics vs. placebo for eradication of PA.
Wiesemann 1998	Ineligible design ‐ participants with new PA infection randomised to 12 months of either tobramycin or placebo, i.e. a prolonged eradication regimen. Although treated group had prolonged treatment there is no comparison group.

IV: intravenous
 PA: Pseudomonas aeruginosa
 vs: versus

Characteristics of studies awaiting assessment [ordered by study ID]

OPTIMIZE 2018.

Methods	RCT. Parallel design. "Double‐blind" for azithromycin vs placebo; the ClinicalTrials.gov entry states the trial was blinded four ways (participant, care provider, investigator and outcomes assessor). Duration: planned 18 months, closed early after a median 11.8 months follow‐up as the trial had reached a pre‐specified interim monitoring boundary for efficacy of the primary endpoint. Multicentre: 45 CF centres in the USA.
Participants	Eligibility criteria Diagnosis of CF and new (defined as lifetime first or at least 2 year absence of PA on culture) isolation of PA within last 40 days. Males and females aged ≧ 6 months and ≦18 months who were clinically stable at time of enrolment. Participants could not have had macrolide antibiotics within 30 days of baseline. Participant characteristics 221 participants randomised; 110 to the active treatment group, 111 to the placebo treatment group. Age, mean (SD) years: active group 7.1 (5.1); placebo group 6.8 (5.0). Gender split, number (%): active group 55 (50.0) females; placebo group 49 (44.9) females. FEV1 % predicted, mean (SD): active group 94.9 (18.0); placebo 93.4 (16.1) group. Weight, mean (SD) kg: active group 26.7 (17.5); placebo group 25.1 (16.2). Mutation, F508del homozygous, number (%): active group 59 (53.6); placebo group 57 (51.4). PA positive at baseline visit, number (%): active group 56 (51.9); placebo group 43 (39.8).
Interventions	All participants received TIS 300 mg 2x daily for 28 days; a further 28 days could be given if a respiratory culture at week 3 was still positive for PA. TIS could have started up to 14 days prior to baseline visit. Active: approximately 10 mg/kg azithromycin suspension up to 500 mg total dose 3x per week. Placebo: volume‐matched placebo suspension. For both groups 2x daily TIS 300 mg quarterly, if respiratory samples in that quarter positive for PA.
Outcomes	Primary clinical outcome Time to first pulmonary exacerbation (using an a priori definition) requiring oral, inhaled or intravenous antibiotics. Secondary outcomes Time to PA recurrence after the first quarter of therapy. Safety (adverse events, ECG and audiologic monitoring). *Frequency of exacerbations. *Frequency of positive PA cultures. *Rate of antibiotic usage. *Rate of hospitalisations. *Change in height. *Change in weight. *Change in lung function (in participants aged 4 or over and able to perform spirometry). *Change in participant/parent reported respiratory symptoms using the Chronic Respiratory Infection. *Symptom score. * Secondary endpoints listed in the paper but not stated in ClinicalTrials.gov protocol
Notes	Unclear whether participants who started TIS prior to enrolment got additional days of treatment.

CF: cystic fibrosis
 ECG: electrocardiogram
 FEV₁: forced expiratory volume in 1 second
 PA: Pseudomonas aeruginosa
 RCT: randomised controlled trial
 SD: standard deviation
 TIS: tobramycin for inhalation solution

Characteristics of ongoing studies [ordered by study ID]

Larsson 2011.

Trial name or title	Phase III study (IMPACTT) on anti‐pseudomonas IgY.
Methods	RCT. Parallel design. "Double‐blind" ‐ participant and investigator. Duration: 2 years. Multicentre: centres in the following European countries: Austria, Belgium, Germany, Hungary. Ireland, Italy, Poland, Spain and Sweden.
Participants	Eligibility criteria Confirmed diagnosis of CF with confirmed eradication of PA in last 3 years and PA negative at screening. Males and females, aged 5 years or over, able to gargle and with FEV1 50% ‐ 130% predicted. Participant characteristics 164 participants enrolled.
Interventions	Active intervention: avian polyclonal anti‐PA antibodies in a 70 mL gargle solution (contains 50 mg IgY) 1x daily. Control intervention: placebo 70 mL gargling solution 1x daily.
Outcomes	Primary outcome Time from start of treatment to the first recurrence of PA in sputum, throat cough swab or endolaryngeal suction. Secondary outcomes Change in FEV1 from day 0 to each visit. Change in BMI from day 0 to each visit. Number of exacerbations. Number of days of illness in hospital and at home (i.e. out of school or work). Control of use of antibiotics measured as days with antibiotic treatment. Change in values of serological tests for PA precipitins from day 0 to each visit (if applicable). Good tolerability and comparable number and quality of adverse events to placebo group. Sputum or throat cough swab or endolaryngeal suction cultures for other bacteria or fungi.
Starting date	October 2011.
Contact information	Dr Jutta Bend Mukoviszidose Institut Gemeinnützige Gesellschaft für Forschung und Therapieentwicklung mbH In den Dauen 6 53117 Bonn Telefon: +49(0)228 987 80 47 Fax: +49 (0)228‐98780‐77 Email: jbend@muko.info
Notes	We have contacted the research team who expect to have a publication ready in 2019.

BMI: body mass index
 CF: cystic fibrosis
 FEV₁: forced expiratory volume in 1 second
 PA: Pseudomonas aeruginosa
 RCT: randomised controlled trial

Differences between protocol and review

We changed out time‐points to include 'up‐to' each pre‐specified time point to include data from trials where data were not available at the pre‐specified time‐points and facilitate comparisons between trials. The exact time‐points reported by each trial were reported in the text.

Where individual patient data was made available for review we have analysed some on an available case‐basis, rather than an intention‐to treat basis.

We planned to define recurrence as a new isolation of Pseudomonas aeruginosa (PA) after a period of six months of negative cultures, where cultures had been taken at least six‐monthly. This was in conflict with our stated primary outcome of time to next isolation of PA. We have therefore analysed on the basis of our primary outcome.

Contributions of authors

Roles and responsibilities
TASK	WHO WILL UNDERTAKE THE TASK?
Protocol stage: draft the protocol	Dr Sally Palser Dr Edward Nash Prof Alan Smyth Dr Arnav Agarwal
Review stage: select which trials to include (2 + 1 arbiter)	Dr Sally Palser Dr Arnav Agarwal Mrs Sherie Smith
Review stage: extract data from trials (2 people)	Dr Sally Palser Mrs Sherie Smith
Review stage: enter data into RevMan	Dr Sally Palser
Review stage: carry out the analysis	Dr Sally Palser
Review stage: interpret the analysis	Dr Sally Palser Dr Edward Nash Prof Alan Smyth Mrs Sherie Smith
Review stage: draft the final review	Dr Sally Palser Dr Edward Nash Prof Alan Smyth Sherie Smith
Update stage: update the review	Dr Sally Palser Dr Edward Nash

Sources of support

Internal sources

• The University of Nottingham, UK.

  Funding for Professor Smyth

External sources

• National Institute of Health Research, UK.

  Funding for Dr S. Palser is from an NIHR Research for Patient Benefit project grant, PB‐PG‐0213‐30055.

• National Institute for Health Research, UK.

  This systematic review was supported by the National Institute for Health Research, via Cochrane Infrastructure funding to the Cochrane Cystic Fibrosis and Genetic Disorders Group.

Declarations of interest

Dr Sally Palser declares that her salary is funded by a Research for Patient Benefit grant from the National Institute of Health Research (NIHR), PB‐PG‐0213‐30055.

Dr Edward Nash declares no known potential conflict of interest.

Dr Arnav Agarwal declares no known potential conflict of interest.

Sherie Smith declares no known potential conflict of interest.

Prof Alan Smyth declares relevant activities of consultancy and an Investigator Initiated Award (independent research grant) from Vertex Pharma. He has given lectures at symposia sponsored by Novartis and TEVA. In addition, Prof Smyth has a patent ALKYL QUINOLONES AS BIOMARKERS OF PSEUDOMONAS AERUGINOSA INFECTION AND USES THEREOF issued.

New