Background: Chlorhexidine oral rinses decrease ventilator-associated pneumonia but its effect on COPD is unknown.

Research Question: Will a preliminary study of the effect of twice-daily chlorhexidine oral rinses on oral and lung microbiota biomass and respiratory symptoms, compared to placebo, support the conduct of a larger clinical trial?

Study Design and Methods: Participants aged 40-85 with COPD and chronic respiratory symptoms were randomized 1:1 to twice-daily 0.12% chlorhexidine oral rinses vs. placebo.

Results: Forty-four participants were recruited between September 8, 2014 and May 30, 2019. Our primary outcome was a change in oral and sputum microbiota biomass during the study as assessed by 16S rRNA copy numbers. Neither the oral microbiota nor the sputum microbiota biomass decreased significantly in those using chlorhexidine compared with placebo (oral microbiota mean log10 difference [SE] = -0.103 [0.23], 95% CI: -0.59, 0.38, p=0.665; sputum microbiota 0.80 [0.46], 95% CI: -0.15, 1.75, p=0.096). Chlorhexidine decreased both oral and sputum microbiota alpha (Shannon) diversity (linear regression estimate [SE] oral: -0.349 [0.091], p=0.001; sputum -0.622 [0.169], p=0.001). Chlorhexidine use did not decrease systemic inflammatory markers compared to placebo (CRP [chlorhexidine 1.8 ± 7.5 vs. placebo 0.4 ± 6.8, p=0.467], fibrinogen [22.5 ± 77.8 vs. 10.0 ± 77.0, p=0.406], or leukocytes [0.2 ± 1.8 vs. 0.5 ± 1.8, p=0.560]). Chlorhexidine use decreased St. George’s Respiratory Questionnaire scores compared to placebo (chlorhexidine -4.7 ± 8.0 vs. placebo 1.7 ± 8.9, p=0.032).

Interpretation: We did not detect a significant difference in microbiota biomass due to chlorhexidine use. Chlorhexidine decreased oral and sputum microbiota alpha diversity and improved respiratory health-related quality of life compared to placebo. Our results support the performance of a larger clinical trial.

At visit 1, participants provided medical history, performed spirometry, completed the St. George’s Respiratory Questionnaire (SGRQ), were instructed on how to complete the Breathlessness, Cough, and Sputum Scale (BCSS) daily diaries, and provided blood, oral, and induced sputum samples prior to randomization. Oral and sputum sample volumes were recorded. Participants returned 8 weeks later to return BCSS diaries, complete the SGRQ, assess outcomes, and provide blood, oral, and sputum samples.

The clinical laboratories at the MVAMC determined WBC and differential, fibrinogen, CRP levels, and sputum gram stain and culture results. All oral rinses, sputum samples, and unused sterile water (control samples) were frozen immediately and until DNA extraction. 16S rRNA quantification and 16S rRNA V4 MiSeq sequencing was performed at the University of Minnesota Genomics Center as previously described.

The primary outcome was change in oral and sputum microbiota biomass after 8 weeks of chlorhexidine vs. placebo use, compared to baseline values as assessed by 16S rRNA quantification. The primary outcome was chosen based on the mechanism of action of chlorhexidine, however sample size calculations were based on a change in alpha diversity (a secondary outcome) due to data availability at study initiation. At a sample size of 20 per group and across a plausible range of effect sizes, our study had 67-94% power to detect a change in alpha diversity associated with chlorhexidine use. Sample size calculations are available in the online supplement. Secondary outcomes included: sputum and oral microbiota Shannon and Simpson diversity; sputum and oral microbiota taxonomy; inflammatory markers (WBC, fibrinogen, and CRP); BCSS scores; SGRQ score; and assessment of adverse events.

Baseline variables were compared using Fisher's Exact Test for categorical variables or the Wilcoxon Two-Sample Test for continuous variables. Means are presented with standard deviations (SD); mean differences and parameter estimates are presented with their associated standard error (SE).

All analyses were performed using SAS version 9.4 (SAS Institute) and the intention-to-treat principle. A two-sided type I error of 0.05 was used. Correction of the Type I error rate for multiple testing was performed using the Step-down Bonferroni method.

For the primary analysis of both normalized oral wash and normalized sputum biomass count, values were transformed to the log₁₀ scale and the mean difference between treatment groups was compared using the two-sample t-test. A multiple imputation procedure was used to impute each unavailable sputum weight.

Linear regression was used to examine the effect of treatment group on the 8-week change in the Shannon and Simpson biodiversity indices, BCSS, SGRQ and inflammatory markers separately, with each model adjusted for the baseline value of the measure.

Subgroup analyses of participants who did not receive antibiotics during the study were also performed for the outcomes of biomass and biodiversity.

For taxa abundance analyses, treatment effects on abundance were examined by modeling the 8-week change using linear regression, adjusted for baseline count. Analysis was restricted to genera with <20% of values equal to zero. Fisher’s Exact Test was used to determine the proportion with a genus detected at Week 8 vs. baseline compared between treatment groups. Results were corrected for multiple comparisons.