Sputum bacterial load and bacterial composition correlate with lung function and are altered by long-term azithromycin treatment in children with HIV-associated chronic lung disease

Table 3 Univariate linear regression analysis of within-participant change in FEV1z and FVCz and within-participant change in relative abundance of selected genera (outcome) between visits

Within-participant change in genus	Trial arm	Within-participant change in FEV1z				Within-participant change in FVCz
Within-participant change in genus	Trial arm	coef	stderr	pval	qval	coef	stderr	pval	qval
Streptococcus	AZM	3.22	1.11	0.004	0.01	2.66	1.04	0.01	0.08
Streptococcus	Placebo	− 0.79	1.46	0.59	0.82	− 0.42	1.2	0.72	0.93
Prevotella	AZM	0.1	0.86	0.91	0.97	0.59	0.78	0.45	0.63
Prevotella	Placebo	− 1.54	1.08	0.16	0.82	− 1.56	0.87	0.07	0.51
Haemophilus	AZM	− 0.92	1.95	0.64	0.89	− 1.12	1.82	0.54	0.63
Haemophilus	Placebo	− 0.76	2.86	0.79	0.9	0.15	2.3	0.95	0.95
Moraxella	AZM	− 2.74	0.74	0.0003	0.002	− 1.28	0.7	0.07	0.25
Moraxella	Placebo	0.18	1.42	0.9	0.9	0.92	1.11	0.41	0.93
Neisseria	AZM	0.05	1.43	0.97	0.97	− 0.34	1.33	0.8	0.8
Neisseria	Placebo	1.5	1.83	0.41	0.82	2.66	1.04	0.01	0.08

Associations were tested with MaAsLin2 using a linear regression model with FEV1z or FVCz and trial arm as fixed effects and within-participant change in relative abundance of selected genera as outcome. Statistical significance was corrected for multiple testing using Benjamini/Hochberg correction. Columns correspond to the within-participant change in genus, trial arm, the coefficient estimate (coef) and standard error from the model (stderr), nominal p-value (pval), and FDR corrected p-value (qval). Number of samples in azithromycin (AZM) and placebo arms are 377 and 365 respectively

ISSN: 2049-2618