Table 2. Multinomial logistic regression of factors associated with having a vaginal microbiome dominated by different bacterial species.

Outcome by dominant bacteria		Relative risk ratio (RRR)	95% CI	p valuea	Adjusted RRR	95% CI	p valueb
L. crispatus vs other non-lactobacilli
	Caucasianc	9.71	1.90, 49.65	0.006	7.28	1.37, 38.57	0.020
	Continuous OCP-used	0.67	0.11, 4.13	0.663	0.14	0.01, 1.31	0.084
Sexual activity category
	Non-coital activities onlye	0.90	0.14, 5.85	0.916	0.51	0.84, 3.04	0.456
	Penile-vaginal sexe	7.86	1.01, 8.21	0.048	11.21	0.93, 134.74	0.057
L. iners vs other non-lactobacilli
	Caucasianc	2.34	0.49, 11.30	0.288	1.71	0.34, 8.53	0.512
	Continuous OCP-used	0.60	0.10, 3.52	0.567	0.16	0.02, 1.10	0.063
Sexual activity category
	Non-coital activities onlye	1.75	0.31, 10.12	0.532	0.94	0.19, 4.76	0.942
	Penile-vaginal sexe	11.00	1.71, 70.73	0.012	17.51	2.18, 140.33	0.007
L. gasseri or L. jenseniif vs other non-lactobacilli
	Caucasianc	2.43	0.36, 16.22	0.360	1.88	0.24, 14.76	0.548
	Continuous OCP-used	0.76	0.10, 5.97	0.792	0.24	0.02, 2.50	0.233
Sexual activity category
	Non-coital activities onlye	0.20	0.02, 1.78	0.149	0.12	0.01, 1.13	0.063
	Penile-vaginal sexe	2.20	0.23, 20.70	0.491	3.00	0.30, 30.16	0.350
G. vaginalis vs other non-lactobacilli
	Caucasianc	1.32	0.20, 8.71	0.770	0.93	0.12, 7.49	0.949
	Continuous OCP-used	0.51	0.05, 5.22	0.573	0.11	0.01, 1.22	0.072
Sexual activity category
	Non-coital activities onlye	0.50	0.08, 3.29	0.471	0.25	0.04, 1.52	0.132
	Penile-vaginal sexe	7.33	0.84, 63.74	0.071	14.03	1.22, 160.69	0.034

^aMultinomial logistic regression with specimens dominated by other heterogeneous non-lactobacilli (predominately A. vaginae and Prevotella spp.) as the referent group, clustered for multiple specimens from participants (52 clusters). Outcome refers to a composite outcome whereby the relative risk of having a specimen dominated by each group were compared simultaneously using multinomial logistic regression models

^bmultinomial logistic regression as before, adjusting for all other characteristics analysed

^crelative to non-Caucasian women which included women from Central or South-East Asia, India or Sri-Lanka

^dspecimens were from women who reported continuous oral contraceptive pill (OCP) use at each interval of follow-up to study endpoint, relative to non-OCP users

^eunprotected penile-vaginal sex relative to no sexual activity

^fspecimens dominated by L. gasseri and L. jensenii were combined for analysis due to smaller sample sizes for both spp.