Criteria for proof of causation

Criterion reported by:
Koch (1890)Rivers (1936)Fredericks and Relman (1996)
A microbe must be present in every case of a diseaseA virus must be associated with a disease with a degree of regularityCandidate sequences should be present in most cases of disease and at sites of disease pathology
It must not be found in association with any other diseaseThe association cannot be incidentalFew or no sequences should be present in host or tissue without disease
After isolation and propagation in pure culture, the microbe must be capable of inducing the same diseaseRivers predicted that methods for proving causal relationships would evolve with improvements in technology; at the time of his writing, he invoked seroconversion and experimental inoculations but acknowledged the limitations of bothSequences should diminish in frequency with resolution of disease and increase with relapse
Sequences should be present prior to the onset of disease