The problem
Changing what someone eats can change the microbes in their gut. If a health measurement changes too, did the microbes cause it, did the food act directly, or did changes in the body alter the microbes? The microbiome is the community of microbes and their genetic capabilities. Which changes in that community actually change the person’s body, and through what process?[1] [2]
A gut microbial profile can change with diet and can differ between people with different health conditions. Does a particular community change cause the host outcome, reflect it, or share another cause with it?
Consider a hypothetical study where a dietary change accompanies both a microbial change and a host measurement. Diet may affect both; the host’s changed environment may also favor different microbes. These alternative causal pathways can fit an association. Statistical causal discovery tries to narrow the possible pathways, but its conclusions depend on assumptions about what was measured and what hidden causes remain.
Interventions supply another kind of evidence. Ridaura and colleagues transferred communities from human twin donors with differing obesity status into germ-free mice. Donor communities transmitted differences in body and fat mass under the studied conditions; diet and the movement of microbes between cohoused mice changed those effects. The experiment establishes microbial contributions in that recipient system, while leaving their human translation and detailed mechanism to test.[1] [2] [3]
Scope
Identifying causal effects and mechanisms of human-associated microbial communities, with explicit attention to confounding, interactions, and host context.
What remains open
Observational causal-discovery methods depend on assumptions, including hidden confounding. They can guide experiments but do not replace them or establish transfer across contexts.[1] [3]
Why it matters
The distinction determines what researchers should try changing. A microbial difference might be an active participant, a response to illness or a marker of the person’s diet. Treating all three as the same explanation can send experiments after the wrong target and encourage unsupported claims about changing the microbiome.[1] [2]
Knowing whether a microbial difference is a cause changes the explanation and the next experiment. A direct microbial effect calls for identifying its products or host pathway; a shared dietary cause calls for a different comparison. Community-transfer experiments also reveal how effects can depend on interacting organisms and food. Combining these distinctions prevents a plausible association from becoming an unsupported claim that adding or removing one microbe will improve health.[1] [2] [3]
Definitions
2 terms
confounding
A shared cause can affect both a microbial measurement and a host outcome, creating an association between them. If that cause is unmeasured or inadequately controlled, the association can be mistaken for a microbial effect.[1]
causal discovery
Trying to identify possible directions of influence among measured variables. Observational data can leave several explanations compatible with the evidence; conclusions depend on assumptions about hidden causes and the model used.[1]
Sources
- Causal discovery for the microbiome
Notes on Causal discovery for the microbiome
Personal View on causal discovery: associations, shared causes and host feedback need explicit assumptions and intervention checks. Its algorithm examples are simulations, not demonstrated human treatment effects.
- Diet rapidly and reproducibly alters the human gut microbiome
Notes on Diet rapidly and reproducibly alters the human gut microbiome
Human short-term diet intervention demonstrates microbial community/activity changes; those changes alone do not identify a particular health benefit.
- Gut microbiota from twins discordant for obesity modulate metabolism in mice
Notes on Gut microbiota from twins discordant for obesity modulate metabolism in mice
Human-donor community transfers affect body and fat mass in germ-free recipient mice; cohousing and diet alter the outcome. This establishes effects in that recipient system, not a human weight-loss treatment.