|Home | About | Journals | Submit | Contact Us | Français|
This European Respiratory Society statement provides a comprehensive overview on protracted bacterial bronchitis (PBB) in children. A task force of experts, consisting of clinicians from Europe and Australia who manage children with PBB determined the overall scope of this statement through consensus. Systematic reviews addressing key questions were undertaken, diagrams in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement constructed and findings of relevant studies summarised. The final content of this statement was agreed upon by all members.
The current knowledge regarding PBB is presented, including the definition, microbiology data, known pathobiology, bronchoalveolar lavage findings and treatment strategies to manage these children. Evidence for the definition of PBB was sought specifically and presented. In addition, the task force identified several major clinical areas in PBB requiring further research, including collecting more prospective data to better identify the disease burden within the community, determining its natural history, a better understanding of the underlying disease mechanisms and how to optimise its treatment, with a particular requirement for randomised controlled trials to be conducted in primary care.
ERS task force reviews evidence and suggests future research for protracted bacterial bronchitis in children http://ow.ly/KDtV30dbmGH
Cough is the most common presenting symptom in those seeking primary healthcare . Within this large group of patients are children similar to the child described in box 1 in having a chronic (>4 weeks) cough, where there may be considerable, but often unrecognised, morbidity adversely impacting upon their quality of life (QoL) [2, 3]. Among children referred to a specialist respiratory clinic because of chronic cough, >80% had sought medical advice on five or more occasions during the preceding 12 months and 53% had been taken to the doctor >10 times during the same period .
A 3-year-old male attends your clinic with his mother who is concerned about his persistent cough and “rattly breathing” that have both been present and unrelenting over the past 6 weeks. Upon taking the history, you learn that his symptoms were accompanied initially by a low-grade fever and nasal discharge, and while these resolved in 2 days and 7 days, respectively, his cough persisted. You identify that the cough is wet in nature. He is well grown, fully immunised and has no history of aspiration or recurrent sinopulmonary infections. There is no family history of chronic pulmonary disease, but the father smokes cigarettes outside the home. The child attends childcare. You confirm on examination the presence of a spontaneous wet cough. There are no signs of upper airway infection and no other abnormal physical findings are present. The chest radiograph shows only perihilar changes.
You consider that this child probably has protracted bacterial bronchitis (PBB). What is the evidence for the existence of PBB? How is it diagnosed and managed? What is its prognosis? What are the causes and risk factors and how can it be prevented?
Protracted bacterial bronchitis (PBB) is characterised by an isolated chronic wet or productive cough without signs of another cause, and usually responds to 2 weeks of an appropriate oral antibiotic. The term was first described as a diagnostic entity by the Brisbane group in 2006  and was recognised then in guidelines as a cause of chronic wet cough in children [5, 6]. Although the true prevalence of PBB within the community is unknown, single and multicentre studies in Australia [4, 7] and Turkey [8, 9] diagnosed PBB in 11–41% of children referred to specialist respiratory clinics, where it was found to be one of the most common causes of chronic cough.
PBB is not a new entity and PBB-like conditions were reported during the last century . In the 1940s the possibility of a link between chronic bronchitis and bronchiectasis was raised , including suggestions that this could be interrupted by intensive antibiotic therapy [11, 12]. Later, in the 1980s, a retrospective review of 20 children with chronic bronchitis reported bronchoscopic evidence of bronchial wall inflammation, purulent bronchial secretions, containing mainly Haemophilus influenzae, and most improved following antibiotic therapy . This report coincided with publication of Cole's  “vicious circle” hypothesis of chronic bacterial infection and inflammation causing bronchiectasis, which helped to provide a conceptual framework for PBB as a potential pre-bronchiectasis state in some children .
PBB is often misdiagnosed as asthma, resulting in inappropriate and often high doses of inhaled corticosteroids [7, 16]. Generic health-related (PedsQL ) and chronic cough-specific (PC-QoL) QoL scores of children with PBB were found to be similar to those recorded in children from other diagnostic groups (asthma, bronchiectasis and those whose chronic cough resolved without treatment) when the QoL measures were undertaken at the first presentation to respiratory specialists [7, 18], but normalised when the cough resolved .
Thus, while PBB-like descriptions are certainly not new, with the clinical features being well described in previous decades, it is now a distinct diagnostic entity and our knowledge of the clinical and pathobiological features has progressed rapidly in recent years. However, as studies of PBB have relied upon children presenting to hospital and specialist clinics, our understanding of the true underlying epidemiology and disease burden of PBB is limited by the absence of community data. This European Respiratory Society (ERS) task force position statement on PBB outlines current knowledge and provides a clinical profile, diagnostic indications and a therapeutic approach to this common disease in children. In addition, it highlights areas of future research.
The ERS PBB task force team consisted of 11 members (general paediatricians and respiratory and infectious disease specialists) representing clinicians in Europe and Australia who manage children with chronic cough. ERS standardised procedures for conflict of interest declaration were followed. The key questions (KQ), framed in a patient intervention comparison outcome format, were developed by the group and distributed among four pairs of authors (online supplementary file 1). Author pairs undertook the systematic reviews based on the method used in the paediatric sections of the American College of Chest Physicians (ACCP) CHEST cough guidelines [19, 20] and the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) statement (online supplementary figures S1–S7).
Symptoms: chest pain, history suggestive of inhaled foreign body, dyspnoea, exertional dyspnoea, haemoptysis, failure to thrive, feeding difficulties (including choking/vomiting), cardiac or neurodevelopmental abnormalities, recurrent sinopulmonary infections, immunodeficiency or epidemiological risk factors for exposure to tuberculosis
Signs: respiratory distress, digital clubbing, chest wall deformity or auscultatory crackles
Tests: chest radiographic changes (other than perihilar changes) or lung function abnormalities
Searches relating to the key questions were undertaken by four pairs of task force members, whereby one member of each reviewer pair used a standard format (online supplementary file 1) updated from and based on the ACCP's CHEST cough paediatric guidelines [19, 20]. The search criteria used for each key question are presented in online supplementary file 1. Both reviewers within each pair independently reviewed all abstracts and agreed upon which full-text articles to retrieve to assess for potentially eligible studies. It was decided that disagreements not resolved by consensus would be adjudicated by a third reviewer (AK). Risk of bias criteria was not undertaken as this was undertaken in the previous systematic review , and no new randomised controlled trials (RCTs) were identified in our searches for this systematic review. For cohort studies, data were extracted by a single reviewer and checked by another reviewer. In cohort studies, we reported on the study's setting, number enrolled and completing the study, inclusion and exclusion criteria and main results relating to the key questions.
Summaries of the data relating to each key question were presented to the entire group at a face-to-face meeting in April 2016. Following a review of the data and discussions, this document and statements were formulated. Consensus was defined a priori as agreement by >80% of the group. Key findings from the key questions and task force statements are presented below with additional details presented in the online supplementary files. Studies had components common to almost all the key questions and they are summarised in table 1.
As outlined recently , the task force considers a reliable definition of PBB for day-to-day clinical practice when all three of the following criteria are fulfilled. 1) Presence of continuous chronic (>4 weeks’ duration) wet or productive cough; 2) absence of symptoms or signs (i.e. specific cough pointers) suggestive of other causes of wet or productive cough (see box 2); and 3) cough resolved following a 2–4-week course of an appropriate oral antibiotic.
The validity of each criterion was outlined in a recent review  and updated in this systematic review for duration of cough, in the absence of other symptoms and signs (online supplementary table S1.1a-b). The task force chose the 4-week diagnostic threshold based upon the inclusion criteria of the studies outlined in online supplementary table S1.1a-b. Other than the British Thoracic Society guidelines, all other national guidelines use a cough duration >4 weeks. However, the duration of cough may need adjusting if new data arise from prospective longitudinal observational studies. This was also recommended by the ACCP cough guidelines . Such information is especially required from within the primary healthcare setting.
In prospective studies that are currently available (online supplementary table S1.1a), the mean or median duration of cough (at presentation) varied from 3 weeks  to as long as 6 months . In the retrospective studies (online supplementary table S1.1b), the mean or median cough duration varied from 10 weeks  to 11 months . The variability in central estimates and 95% confidence intervals of these studies was high, and even more so in the retrospective studies.
PBB as a diagnostic entity and cause of wet cough was well documented in prospective and retrospective studies (online supplementary table S1.1a-b). The response to oral antibiotics was used to help define PBB in almost all studies and national cough guidelines in children. However, the length of antibiotic courses to define PBB varied between studies. In all but one of the prospective studies, the duration of antibiotic treatment ranged from 10 days to 2 weeks (online supplementary table S1.1a). However, in the retrospective studies, the mean duration of antibiotic courses either varied from 17 days to 6–8 weeks or was unspecified (online supplementary table S1.1b).
Prospective studies were supported by other descriptive studies examining the associated bacteriology, inflammatory profiles and immune responses, providing further evidence of PBB as a diagnostic entity in children (online supplementary table S1.3a-b). While an adult version of PBB has not been clearly described, it has been suggested by some clinicians . However, in this particular report , other than presence of wet or productive cough, other criteria such as duration of cough went undefined. Furthermore, another adult case series  described chronic productive cough unrelated to bronchiectasis, which responded only to intravenous antibiotics. However, this requirement for parenteral antibiotics is used to help to differentiate between chronic suppurative lung disease (CSLD) and PBB in children .
Although most current guidelines include PBB as a cause of chronic wet cough, variations exist in the definition of PBB, and some do not clearly define PBB (online supplementary table S1.2). While all definitions include the presence of chronic wet or productive cough, there is variability in the inclusion/exclusion of treatment response, microbial isolation (and method used). Some documents use BAL as a diagnostic criterion; other documents use microbiological findings, yet others just clinical criteria. As recently described , the original microbiology-based case definition  (also termed PBB-micro) criteria are as follows. 1) Presence of chronic wet cough (>4 weeks); 2) lower airway infection (recognised respiratory bacterial pathogens growing in sputum or at bronchoalveolar lavage at density of a single bacterial species >104 CFU·mL-1); and 3) cough resolved following a 2-week course of an appropriate oral antibiotic (usually amoxicillin-clavulanate). PBB-clinical (modified clinical-based case definition [6, 57]) refers to presence of the following criteria. 1) Presence of chronic wet cough (>4 weeks); 2) absence of symptoms or signs of other causes of wet or productive cough; and 3) cough resolved following a 2-week course of an appropriate oral antibiotic (usually amoxicillin-clavulanate).
Part of KQ1 reviewed what symptoms and signs (other than wet cough) were used (or found) in descriptions of children with PBB. The prospective studies (online supplementary table S1.3a) specified that no other symptoms (other than wet cough) or signs were present, although some had “parent-reported wheeze”. The retrospective studies (online supplementary table S1.3b) commonly described wheeze or “noisy breathing” being present. Concomitant bronchodilator responsiveness was described in some studies, and was particularly common in one retrospective study (nine out of 20 children ), although this high proportion could have arisen through patient selection.
As summarised previously , while children with PBB were typically young (median age ranging from 10 months to 4.8 years), PBB was also recognised in older (age >12 years) children [7, 37, 51]. Most children with PBB lacked systemic symptoms, without evidence of sinusitis or ear disease [4, 43]. Compared to “disease controls” undergoing bronchoscopy for non-cough-related indications (e.g. stridor or apnoea), children with PBB were more likely to have attended childcare (OR 8.4, 95% CI 2.3–30.5), but their exposure to tobacco smoke (~30% [4, 43]) was similar to that of the “controls” . Children with PBB typically appeared well. They had normal growth and development, and lacked signs of underlying CSLD, such as digital clubbing, chest wall deformity or adventitial auscultatory chest findings , although occasionally a “rattly chest” and crackles were heard.
The prevalence of atopic features (eczema, systemic and airway eosinophilia, elevated IgE or positive radioallergosorbent test) was similar to children without PBB . While many parents (41–81%) reported previous “ever wheeze” [16, 43], auscultation-confirmed wheeze by doctors was unusual.
The chest radiograph was normal or near-normal, showing only peribronchial changes [15, 28, 60]. When performed, both spirometry  and respiratory system reactance were also normal. While no RCTs have assessed whether ordering chest radiographs and/or spirometry in children with PBB or chronic wet cough enhanced management, a systematic review found “high-quality” evidence for improved clinical outcomes by adopting cough management protocols (or algorithms) in children with chronic cough aged <14 years . Steps in these cough management algorithms included undertaking a chest radiograph and (when possible) spirometry . When an abnormality in either the spirometry and/or the chest radiograph (other than peribronchial changes) in a child with chronic wet cough is present, additional investigations for an underlying cause are indicated (figure 1). PBB may co-exist with other diseases, (e.g. asthma and airway malacia), although for asthma, no studies including objective assessments of reversible airflow limitation to help determine whether there is any association have been undertaken.
Data from KQ2 included evaluating causes of chronic wet or productive cough in children when they first present to doctors (online supplementary table S2.1). These other causes of chronic wet cough in children include, but are not limited to pertussis, tuberculosis, inhaled foreign body, bronchiectasis, cystic fibrosis, aspiration or congenital lung lesions. Most however have other symptoms and signs present (i.e. cough pointers [19, 28, 60]).
In addition, KQ2 addressed other investigations and the possible harm from having a prolonged wet cough (online supplementary tables S2.2 and 2.3). Data suggest that children should be referred for further investigations when specific cough pointers (box 2) are present or when the wet cough does not respond to 4 weeks of antibiotics. One study found that failure of the cough to respond to 4 weeks of antibiotics increased the chance of bronchiectasis being present (adjusted OR 20.9, 95% CI 5.4–81.8) . Another study reported that the duration of chronic wet cough was significantly associated with increased risk of structural airway abnormalities and increased Bhalla scores on chest computed tomography (CT) scans .
The respiratory bacterial pathogens found in studies of chronic wet cough are summarised in online supplementary table S3.1. Overall, H. influenzae was the most common organism, found in 28–58% of children [4, 31, 32, 37, 42, 56], with Streptococcus pneumoniae (13–58%) and Moraxella catarrhalis (17–59%) the two other most frequently detected organisms. These results are similar to findings in PBB (online supplementary table S3.2).
The first study describing PBB detected commonly recognised respiratory pathogens (H. influenzae (47%), Strep. pneumoniae (35%) and M. catarrhalis (26%)) in BAL cultures at high bacterial loads (≥105 CFU·mL-1) . Subsequently, other studies have supported these findings (online supplementary table S3.2). As in children with chronic wet cough, H. influenzae was the most common pathogen (38–81%) cultured from children with PBB. Although most H. influenzae are likely to be nontypeable (NTHi) strains, only two studies have attempted capsular typing of H. influenzae isolates [51, 61]. The other commonly detected bacteria are Strep. pneumoniae (16–39%) and M. catarrhalis (19–51%), while Staphylococcus aureus was found (6–22%) in five of the 10 published studies (online supplementary table S3.2). Differences in the types of Strep. pneumoniae serotypes encountered in children with PBB from different countries have been reported and may have arisen from variations in antibiotic prescribing and vaccine practices. For example, in one comparative study, 100% of Strep. pneumoniae isolates from Greek children undergoing BAL for PBB were serotypes contained in the 13-valent pneumococcal conjugate vaccine, while only 28% of Strep. pneumoniae isolates in BAL cultures from fully vaccinated children with PBB in the UK were vaccine-type serotypes . Finally, polymicrobial infections involving multiple respiratory bacterial pathogens were identified in the lower airways of 30–50% of affected children [16, 42, 50–52].
When assessing the bacteriology of PBB it is important to note that not all BAL studies are directly comparable with one another, as only some utilised quantitative bacterial culture to describe BAL findings [50, 51]. The distribution of bacteria within the lungs is not uniform. One retrospective study reported that sampling from a single lobe would have missed 17 different organisms in 15 out of 50 patients, eight of whom would not have had any organisms cultured . Nonetheless, it should be noted that this study used nonquantitative bacterial culture methods and failed to report bacterial densities between the various lobes. A summary of information on bacteria found in PBB is presented in online supplementary table S3.2.
Our systematic review (online supplementary table S4.1) revealed two studies that reported on viruses in children with PBB [43, 44] and another in children with wet cough . However, we found only one study that specifically systematically examined for viruses . The presence of any virus was found in 38% of BAL samples from 104 children with PBB , which was significantly greater than in 49 children with other chronic respiratory disorders (38% versus 9%; OR 6.3, 95% CI 2.1–19.1). The most common virus detected in children with PBB was human adenovirus (HAdV) (23%) of which most cases were representative of the HAdV-C species (detected in 96% of HAdV+ children) . An extended molecular-based diagnostic panel for 17 respiratory viruses was performed in a subset of 27 children, which found rhinovirus in 11 (41%) subjects and human bocavirus and human coronavirus each in one (4%) participant . However, the prevalence of these other viruses was similar in the control group . The same study showed that lower airway infection with H. influenzae (≥104 CFU·mL−1) was significantly associated with HAdV co-infection .
Our systematic review identified a single study that examined the lower airway microbiota of children with PBB . It compared the microbiota of 12 children with PBB with that of 19 children with bronchiectasis and 25 with cystic fibrosis . Unlike adult studies, where significant interindividual differences exist in bacterial community composition between patients with cystic fibrosis and bronchiectasis, the core microbiota in the three aforementioned conditions was similar during early childhood, with H. influenzae and oral aerobic (Streptococcus mitis) and anaerobic (Prevotella melaninogenica) commensals the most commonly shared core species. Subsequent to our systematic review, a larger study involving 78 children (PBB n=28, CSLD/bronchiectasis n=40, disease controls n=10), found that the microbiota in upper and lower airways helped to discriminate between these clinically defined groups . Statistically significant differences have been found in the microbiota between these diagnostic groups .
Besides microbiology, the pathobiology of PBB (addressed in KQ5 and summarised in online supplementary tables S5.1 and 5.2) involves risk factors, underlying structural airway lesions and host airway inflammatory and immune responses. However, few studies have described PBB risk factors, and these report a male predominance [42, 55] and a median age ranging from 10 months to 4.8 years [4, 42, 55]. In addition, a single prospective study found that childcare attendance was the single positive risk factor in 91% of PBB cases, compared with 58% of disease controls .
Studies included in the systematic review describing the BAL inflammatory profile of PBB had similar findings of intense airway neutrophilia (summarised in online supplementary tables S5.1a and S5.1b), with a percentage neutrophil median range of 25.5–44% [4, 16, 23, 26, 36, 38, 39]. The airway neutrophilia was accompanied by a raised total cell count with a median range of 188–426×106 cells·L-1 [4, 23, 26, 36, 38, 39]. No airway eosinophilia was observed in any study and a single study described an increase in the percentage of lymphocytes .
An associated marked inflammatory mediator response was also found in the BAL fluid of children with PBB. Increased levels of interleukin (IL)-8, active matrix metalloproteinase-9 and IL-1β correlated with the degree of neutrophilia [35, 36, 38]. Other pro-inflammatory mediators detected at increased levels in the BAL fluid of PBB patients included α-defensin, IL-1 pathway cytokines, and CXCR2 gene and protein expression [23, 34, 63]. In addition, IL-1β and related mediators were associated with BAL neutrophils, cough symptoms and disease recurrence .
Studies outlined in online supplementary table S5.1b consistently demonstrate that children with PBB have preserved systemic adaptive immunity with normal serum immunoglobulin levels (IgA, IgM, IgG and IgE) and normal antibody-mediated responses to both protein (tetanus) and conjugated protein-polysaccharide based (H. influenzae type b) vaccines [4, 43]. Lymphocyte subsets were normal, except for increased CD56 and CD16 natural killer cell levels for age [43, 55].
Several studies described activated pulmonary innate immune pathways [26, 38, 43, 55, 63]. Specifically, the toll-like receptors (TRLs) associated with bacterial infection (TLR-2 and -4) were elevated in PBB. Another study reported augmented human β-defensin-2 and mannose-binding lectin levels, while activated caspase-1 dependent pro-inflammatory pathways in response to NTHi were also detected in children with PBB, indicating that both innate pathogen recognition and clearance mechanisms were intact .
Finally, a recent study of BAL fluid samples from children with PBB discovered reduced alveolar macrophage phagocytic host responses to NTHi and to apoptotic cells (i.e. deficient efferocytosis) . It is therefore possible that the combination of reduced efferocytosis and increased IL-1β pathways could lead to persistence of the activated M1 macrophage phenotype with its pro-inflammatory effects and resulting chronic neutrophilic airway inflammation . Efferocytosis values in children with PBB were in between controls and children with bronchiectasis .
Purulent airway secretions and large airway malacia are common bronchoscopic findings in children with PBB [15, 56, 57]. Studies seeking specifically both PBB and tracheo-bronchomalacia reported airway malacia in 74% of PBB cases in one retrospective study  and 68% in a prospective study , although in the latter 53% of disease controls also had evidence of airway malacia . Systematic review of the available studies shows that it remains unclear whether one condition is an antecedent of the other. While the reduced airway clearance found in tracheo-bronchomalacia is thought to predispose to PBB, it is also possible that chronic infection and airway inflammation leads to the development of secondary airway malacia . A recent study comparing clinical findings, airway cellularity and bacterial cultures in PBB patients with and without tracheo-bronchomalacia observed no differences between the two groups . Hence, although it is a common finding, the role of airway malacia in the pathobiology of PBB remains largely unknown.
13 studies were included in the systematic review for KQ6 and the primary studies are summarised in online supplementary table S6. One was a Cochrane review , three were RCTs [30, 33, 39] and the remaining nine were descriptive studies, of which four involved prospective cohorts of children with a chronic cough [4, 8, 9, 37] and five were retrospective reviews [13, 16, 47, 48, 52].
The three RCTs [30, 33, 39] were the only studies whose principal aims were to determine the efficacy of antibiotics in resolving a chronic wet cough in young children. The Cochrane review included two of these studies, one of which was open-label , and when both were combined they involved 140 children aged ≤7 years [28, 33]. While concluding that antibiotics were likely to be beneficial (pooled OR 0.13, 95% CI 0.06–0.32) for children with persistent cough post-treatment), the review also raised concerns over study quality and design . Both studies relied upon nasopharyngeal cultures to determine the bacterial aetiology of the suspected lower respiratory tract infection and included children with pertussis and cough duration as brief as 10 days, while outcomes were based upon physician's assessment and not defined fully. Antibiotic courses were for 7 days and included erythromycin in one trial  and relatively low doses of amoxicillin-clavulanate in the other . One RCT had a more robust design and found that children treated with conventional doses of amoxicillin-clavulanate for 2 weeks had higher rates of cough resolution than those receiving placebo (48% versus 16%) . However, this trial included only 25 subjects in each treatment arm and was conducted in a specialist clinic, raising the possibility of sample bias towards more severe cases. Importantly, none of the studies undertook long-term follow-up to determine recurrence rates following treatment.
The remaining evidence supporting the role of antibiotics was limited to a small number of observational studies, many involving children with chronic cough from various causes, where controls were absent and cough resolution following antibiotics was not a main outcome (online supplementary table S6). These studies suggested that a ≥2-week course of antibiotics active against respiratory bacterial pathogens (H. influenzae, Strep. pneumoniae, M. catarrhalis and Staph. aureus) found in the lower airways of children with chronic wet or productive cough were associated with increased likelihood of cough resolution [16, 48, 52]. However, symptomatic recurrences were common [16, 48, 52], occurring in as many as 76% of cases, while a poor response to ≥4 weeks of treatment increased the likelihood of the presence of underlying bronchiectasis .
While there was some evidence that antibiotics improved clinical outcomes in children with chronic wet or productive cough unrelated to underlying disease and without any specific cough pointers, larger RCTs of well-characterised children with chronic wet cough recruited from multiple centres to reduce sample bias and with prolonged follow-up are needed. The lack of comparative studies means that important knowledge gaps exist involving the choice of antibiotic, the doses to be prescribed and the treatment duration needed, to optimise the clinical response and reduce the risk of recurrent chronic wet cough.
A recent review recommended that until further evidence is available, children with an isolated chronic wet cough, lacking symptoms and signs of an underlying disease and whose chest radiograph is normal (or shows only peribronchial changes) should receive 2–4 weeks of an oral antibiotic directed against common respiratory bacterial pathogens associated with PBB . As most children are too young to expectorate and provide a reliable spontaneous or induced sputum specimen, the choice of antibiotic is frequently empirical and determined by local patterns of antibiotic susceptibility. The antibiotic used most widely is oral amoxicillin-clavulanate, which is active against β-lactamase-producing strains of H. influenzae, M. catarrhalis and Staph. aureus, although alternatives such as oral second- or third-generation cephalosporins, trimethoprim-sulfamethoxazole or a macrolide may be used when there is a history of an IgE-mediated reaction to penicillin. However, in those with a previous IgE-mediated reaction to penicillin, oral cephalosporins with similar side chains to the implicated penicillin agent (e.g. ampicillin and cefalexin or cefaclor) should be avoided [65, 66]. In addition, this review recommended that investigations for an underlying cause should be undertaken in the small group of children who, despite adhering to their treatment, fail the 4-week course of antibiotic therapy [47, 57].
There are many causes of chronic wet cough in children, of which PBB is just one . In those in whom therapy fails (figure 1), possible reasons are nonadherence and/or other causes of chronic wet cough. Guidelines suggest that these children should be referred for further evaluation and investigations recommended for suspected CSLD or bronchiectasis . Intravenous antibiotics are probably beneficial in those with persistent endobronchial infection, even without CT scan evidence of bronchiectasis , since early intervention to break the “vicious cycle” of infection, inflammation and impaired mucociliary clearance might prevent future development of bronchiectasis [69, 70].
A prolonged period of observation is required to determine the recurrence rate, and only limited data were available, all from three retrospective studies [16, 48, 52], which were included in the systematic review for KQ7. These studies provided little or no evidence for the role of prophylactic antibiotics, risk of increased antibiotic resistance following treatment or how recurrences of PBB should be managed (online supplementary table S7).
The first  reported that while 51% of 81 children with newly diagnosed PBB were completely symptom free after two prolonged (6–8 week) courses of antibiotics, 13% required either six or more courses of antibiotics or had continuous prophylactic antibiotics for at least one winter to control symptoms, with another 5% requiring intermittent antibiotic courses and were still under active review. However, this study  included children with bronchiectasis and used different definition of PBB. The second  noted that over a 2-year period, only 25% of 33 patients remained well following their first 6–8-week course of antibiotics, despite all achieving complete resolution of their cough symptoms by day 14 of therapy. Moreover, three (10%) of these children had three or more recurrences, and overall nine (27%) received long-term prophylactic antibiotics. Like the first study, the definition of PBB was not the same as that proposed in this task force document. Finally, a third retrospective study  reported complete resolution of symptoms in all but one of 61 children who underwent BAL for chronic wet cough followed by ≥2-week course of oral antibiotics. Of these children, 43 (70%) required repeated courses of treatment for recurrent symptoms.
It should be noted there was no systematic approach to starting prophylactic antibiotics, with some children beginning these after a single recurrence, while details on antibiotics, repeat treatment courses and outcome are scant. Microbiology reporting was incomplete and no antibiotic resistance data were presented. Thus, crucial knowledge gaps remain on how to best manage children with recurrent PBB, how these episodes might be prevented and the impact of long-term antibiotics upon antimicrobial resistance in this patient population.
Subsequent to the task force group meeting, the sole published long-term prospective follow-up study described a subgroup of children with PBB who subsequently received a diagnosis of bronchiectasis . Multivariate logistic regression showed that recurrent episodes (more than three per year) of PBB (adjusted (a)OR 11.5, 95% CI 2.3–56.5; p=0.003) and presence of H. influenzae lower airway infection (aOR 7.6, 95% CI 1.5–37.8; p=0.013) were both independently associated with bronchiectasis diagnosis within the 2-year follow-up period .
Whether symptom duration prior to treatment impacts on clinical outcomes remains unknown, as the current definition includes everyone with a persistent wet cough, irrespective of whether it has been present for weeks, months or even years. The risk of recurrence is likely to depend upon factors that are difficult to assess, such as the composition of the respiratory microbiota, the relevance of airway malacia and/or subtle immune deficiencies. In addition, it is also unknown whether the approach to initial therapy (i.e. length of antibiotic treatment course) influences the risk of recurrence. Prescribing 2–4 weeks of oral antibiotic therapy targeted at the suspected (or cultured) bacterial pathogens has been part of the diagnostic process. Should the cough resolve completely, some clinicians will cease therapy at this point, while others will continue for prolonged periods. The proposed rationale for prolonged antibiotic use is that protecting the airways against the common respiratory bacterial pathogens for an extended period allows the airways to recover their integrity and resist reoccurrence. However, this practice is unproven at either a mechanistic or clinical level. In addition, prolonged antibiotic exposure disrupts resident microbiota (dysbiosis) and contributes to the pathoadaption of organisms within the lower airways, including the selection of antibiotic-resistant strains [71, 72].
Finally, while once-weekly azithromycin halved the rate of pulmonary exacerbations in children with either CSLD or bronchiectasis , its role in PBB remains undefined. This is especially important in the setting of the global antibiotic resistance crisis from over-prescribing antibiotics , where, in addition to the concerns stated above, macrolides in particular alter the host's resident microbiome  and provide strong selection pressure for antibiotic-resistant organisms, leading to increased treatment costs and risk of treatment failure [76, 77].
To conclude, we return to the 3-year-old boy who received a 2-week course of amoxicillin-clavulanate, following which his cough resolved completely. His parents were informed of his diagnosis of PBB and counselled that PBB is an endobronchial infection. If a flexible bronchoscopy and lavage were undertaken, we would probably find common respiratory bacteria and airway inflammation. However, we do not really know why PBB occurs in some and not in others. While we now understand some of the associations with PBB and have found that PBB can recur, we do not know who are more likely to experience recurrence or whether an even longer course of antibiotics reduces the chance of having a recurrence. Recurrence of PBB should be managed with further antibiotic courses and if more than three episodes occur per year, further investigations are indicated.
Please note: supplementary material is not edited by the Editorial Office, and is uploaded as it has been supplied by the author.
Supplementary files ERJ-02139-2016_Supplementary_Files
Supplementary figures ERJ-02139-2016_Supplementary_Figures
Supplementary tables ERJ-02139-2016_Supplementary_Tables
This article has supplementary material available from erj.ersjournals.com
Support statement: The task force was supported by the European Respiratory Society (task force 2015-03). Funding information for this article has been deposited with the Crossref Funder Registry.
This document was endorsed by the ERS Executive Committee in June 2017.
Conflict of interest: Disclosures can be found alongside this article at erj.ersjournals.com