Table 2.
Logistic mixed model regression analysis for examining the effect of probiotic supplementation on bacterial genera abundance
| Probiotic (n = 38 infants, 124 specimens) |
Placebo (n = 28 infants, 91 specimens) |
||||||
|---|---|---|---|---|---|---|---|
| Genusa | Prevalence n infants (%)b; n specimens (%)c |
Relative abundance Mean % (SD) |
Prevalence n infants (%)b; n specimens (%)c |
Relative abundance Mean % (SD) |
AORd (95% CI) | P value | Adjusted P valuee |
| Bifidobacterium | 38 (100); 111 (90) | 36.4 (32.5) | 25 (89); 50 (55) | 17.5 (27.4) | 4.28 (2.02–9.10) | < 0.001 | 0.002 |
| Enterobacter | 37 (97); 95 (77) | 14.8 (24.6) | 28 (100); 77 (85) | 18.7 (25.1) | 0.75 (0.40–1.41) | 0.37 | 0.45 |
| Escherichia/Shigella | 37 (97); 84 (68) | 9.1 (19.1) | 28 (100); 74 (81) | 7.5 (17.2) | 0.69 (0.34–1.43) | 0.32 | 0.43 |
| Staphylococcus | 36 (95); 65 (52) | 8.1 (24.0) | 24 (86); 48 (53) | 7.5 (20.4) | 0.85 (0.47–1.55) | 0.60 | 0.60 |
| Enterococcus | 37 (97); 76 (61) | 3.7 (11.0) | 28 (100); 71 (78) | 8.8 (19.3) | 0.37 (0.20–0.71) | 0.003 | 0.02 |
| Streptococcus | 38 (100); 93 (75) | 4.7 (15.0) | 25 (89); 48 (53) | 2.9 (9.4) | 1.57 (0.88–2.80) | 0.13 | 0.31 |
| Veillonella | 32 (84); 56 (45) | 2.4 (7.2) | 25 (89); 46 (51) | 3.3 (7.9) | 0.74 (0.42–1.33) | 0.32 | 0.43 |
| Clostridium | 32 (84); 53 (43) | 0.5 (2.4) | 27 (96); 45 (49) | 3.0 (9.0) | 0.68 (0.38–1.21) | 0.19 | 0.31 |
| Lactobacillus | 23 (61); 41 (33) | 0.9 (3.7) | 13 (46); 19 (21) | 2 (7.6) | 1.85 (0.78–4.41) | 0.16 | 0.31 |
| Citrobacter | 20 (53); 27 (22) | 0.8 (3.4) | 21 (75); 34 (37) | 1.7 (10.1) | 0.44 (0.19–1.02) | 0.06 | 0.21 |
| Akkermansia | 6 (16); 6 (5) | 0.1 (0.5) | 3 (11); 3 (3) | 1.4 (9.2) | 1.87 (0.41–8.59) | 0.42 | 0.47 |
| Pantoea | 11 (29); 14 (11) | 0.1 (0.8) | 12 (43); 18 (20) | 1.1 (10.1) | 0.45 (0.15–1.35) | 0.15 | 0.31 |
Abbreviations: AOR adjusted odds ratio, CI confidence interval, IQR interquartile range; P values <0.05 are bolded to indicate statistically significant associations.
aProportional abundances of each genera were converted to a binary variable (based on the median value). Only genera that had a mean abundance of at least 1% abundant in one (or both) allocation group were included in regression analysis; b Presents the number (and percent) of infants who had at least on specimen over the study period test positive for genus; c Presents the number (and percent) of total specimens test positive for genus d Odds ratio for mixed effects regression model association between allocation group and bacterial abundance adjusted for gestation and age at sampling, clustering by participant number to account for multiple specimens from infants (66 clusters). e P-value false discovery rate adjustment for multiple testing was performed using the Benjamini-Hochberg method