Comparison of cough plates and cough swabs for detecting respiratory pathogens in non-expectorating children with cystic fibrosis

Abstract

Background
Cystic fibrosis (CF) is an autosomal recessive, multisystem disorder affecting ion transportation across epithelial membranes. Respiratory complications are the leading cause of morbidity and mortality and CF lung disease is characterised by repeated infections leading to inflammation and fibrosis of the lungs. Early and accurate identification of respiratory pathogens is essential to enable timely and efficacious treatment of infections. This is ensured by obtaining microbiology samples at each clinic visit. The gold standard is sputum culture; however, many children are unable or unwilling to expectorate. In non expectorating children, cough swabs (CS) or cough plates (CP) can be used. Previous studies comparing the performance of these two samples are conflicting.
Aims
The primary aim of this study is to compare yield of respiratory pathogens identified by CS and CP in non expectorating children with CF. Secondary objectives include comparing the types of organisms identified by CS and CP and assessing whether lung function, Body mass index (BMI) centile, age and gender affect whether a CS or CP is more likely to identify a pathogen.
Methods
Non-expectorating children with CF attending Shrewsbury and Telford NHS Trust provide CS and CP samples at each outpatient visit. The results of the paired cultures from November 2013-2018 were analysed. Samples positive for non pathogenic organisms including yeast, candida and Bacillus species were excluded.
Results
We identified 663 paired CS and CP samples from 38 patients. Mean (SD) age was 9.9 (4.7) years. The CS and/or the CP was positive for a respiratory pathogen on 118 (18%) of the paired samples. This included 66 (10%) CS and 87 (13%) CP samples. Only 27 (23%) of the paired samples with a positive culture identified the same pathogen(s) on both CS and CP. Number needed to sample: in this sample 13 children needed both CS and CP samples to identify one additional respiratory pathogen compared to CS alone. McNemar’s test (p=0.028) and the generalised estimating equation (p=0.020) suggest that the difference in CS and CP performance was statistically significant. However, the magnitude of this difference was so small whether it is clinically significant remains to be determined. CS identified a greater variety of pathogens compared to CP, identifying 6 additional pathogens not cultured on CP. Lung function was the only significant predictor of both CS and CP positivity on multivariable analysis. BMI centile also had a statistically significant relationship with CP being positive for pathogens on multivariable analysis.
Conclusions
The rate of respiratory pathogen isolation was similar for CS and CP samples. Simultaneous use of CP and CS identified more pathogens than either sampling method alone. Further studies are required to investigate if similar improvements could be achieved by obtaining duplicate CS or CP samples.