Table 2.

Hill Criteria for Causation [115].

Criteria and explanation as described in Bradford Hill’s original 1965 paper (153)	Evidence from studies using classical methods for BV diagnosis	Evidence from newer studies using molecular methods to characterize vaginal microbiome	Data satisfy criterion1
Strength: “First upon my list I would put the strength of association […] In thus putting emphasis upon the strength of association we must, nevertheless, look at the obverse of the coin. We must not be too ready to dismiss a cause-and-effect hypothesis merely on the grounds that the observed association appears to be slight.”	Multiple prospective studies and meta-analyses generally agree that both BV and intermediate microbiota on vaginal Gram stain are associated with approximately 1.5-fold higher risk of HIV acquisition compared to normal microbiota.	Diverse non-Lactobacillus dominant vaginal bacterial communities, relative abundance of some bacteria, and quantities of some bacteria have been associated with increased risk of HIV acquisition. Risk estimates vary across the three available studies.	Partially
Consistency: “Has it been repeatedly observed by different persons, in different places, circumstances and times?”	Of 15 prospective studies and meta-analyses, all but 1 found a statistically significant association between BV and HIV acquisition	Across all three studies, more diverse bacterial communities were associated with HIV acquisition. Relationships between individual bacterial taxa and HIV acquisition are more consistent for some species (e.g. Sneathia and Mycoplasma) than for others (e.g. Prevotella, non-iners Lactobacillus species).	Mostly
Specificity: “One reason, needless to say, is the specificity of the association […] If as here, the association is limited to specific workers and to particular sites and types of disease and there is not association between the work and other modes of dying, then clearly that is a strong argument in favour of causation.”	Bacterial vaginosis is not a necessary exposure to acquire HIV.	There is limited evidence, but it seems highly unlikely that a particular vaginal community or the presence or high concentrations of a specific type of bacteria is required for HIV acquisition.	Not at all
Temporality: “My fourth characteristics is the temporal relationship of the association—which is the cart and which the horse?”	Numerous prospective studies have demonstrated an association between BV and subsequent acquisition of HIV.	The three available molecular studies all captured data on the vaginal microbiome prior to or very shortly after, HIV infection.	Fully
Biological gradient: “…if the association is one which can reveal a biological gradient, or dose-response curve, then we should look most carefully for such evidence.”	Prospective studies provide inconsistent evidence of a biological gradient for increased HIV risk with increasing Nugent score or numbers of Amsel criteria fulfilled.	There is some evidence from recent studies of increasing HIV risk with increasing vaginal bacterial community diversity, and with higher relative abundance and absolute concentrations of some types of bacteria	Partially
Plausibility: “It will be helpful if the causation we suspect is biologically plausible. But this is a feature I am convinced we cannot demand.”	Multiple biologically plausible mechanisms could explain the relationship between BV and increased susceptibility to HIV.	Studies at the molecular level suggest that more diverse vaginal bacterial communities, and the presence and concentrations of key bacteria, may increase HIV susceptibility.	Mostly
Coherence: “[…] the cause-and-effect interpretation of our data should not seriously conflict with the generally known facts of the natural history and biology of the disease.”	Laboratory studies have mostly shown plausible mechanisms, as noted in the previous row. However, there is not a good animal model of the Lactobacillus-dominated healthy human microbiota associated with lower risk. Ex vivo infection of biopsy specimens from women with different vaginal conditions could be helpful in providing additional data showing coherence between laboratory data and epidemiological studies.		Minimally
Experiment: “Occasionally it is possible to appeal to experimental, or semi-experimental, evidence. For example, because of an observed association some preventive action is taken. Does it in fact prevent?”	Clinical trials of BV interventions have produced moderate reductions in BV prevalence over periods up to one year. However, no effective interventions for reducing BV have been evaluated in HIV prevention trials.	One study has demonstrated that periodic presumptive treatment with metronidazole plus miconazole reduces detection and concentrations of several BV-related bacteria. However, no effective interventions targeting the vaginal microbiome have been evaluated in HIV prevention trials.	Not at all
Analogy: “In some circumstances it would be fair to judge by analogy.”	Prospective studies and clinical trials provide some evidence for an analogous relationship between BV and STIs other than HIV. However, these are also not proven causal relationships, and could be mediated through separate mechanisms.	There is some evidence that the presence of key bacteria may increase susceptibility to STIs other than HIV. However, these are also not proven causal relationships, and could be mediated through separate mechanisms.	Minimally

Abbreviations: BV=bacterial vaginosis; CST=community state types; HIV=human immunodeficiency virus; STIs=sexually transmitted infections

¹Overall degree to which the available evidence satisfies each of the 9 Hill criteria on a scale that includes not at all, minimally, partially, mostly, fully.