Table 1

Application of Hill's viewpoints on the causal association between the Dent-O-Sept swab and becoming colonised or infected during the outbreak
Hill's viewpoints	Application on the Dent-O-Sept outbreak

1. Strength of association	Strong.
	Association for having used the swab during hospitalisation and having the outbreak strain of P. aeruginosa, adjusted odds ratio 5.3.
	Detecting genotypically identical strains of the bacterium in patients, the product and in the production facility [1]
2. Consistency of association	Yes, to a large extent.
	However, other co-factors also needed to be in place, e.g. contamination of the particular swab and a susceptible patient. Due to secondary spread in the hospitals also patients who did not use the swab were infected.
3. Specificity of association	Yes, mostly.
	Use of contaminated swabs led to colonisation and some times to infection. Necessary co-factors were as above (2). The clinical manifestations of the P. aeruginosa infection varied widely.
4. Temporal sequence of association	Yes.
	However, the outbreak strain of the bacteria was found in six patients before the production of the first contaminated batch of swabs was detected [2].
	When the swabs were withdrawn from the marked the number of cases gradually diminished and disappeared.
5. Biological gradient	This was not tested but assumed. Reuse of the swabs may have increased the bacterial load and hence the risk of becoming infected.
6. Plausibility of association	Yes.
	The chain from contamination during production to infection is well described.
7. Coherence of association	Yes.
	There is no other hypothesis of explanations for the outbreak.
8. Experiment (reversibility)	Yes, a natural experiment.
	When the source was removed the number of cases gradually diminished to zero.
9. Analogy	Yes.
	There are many other outbreaks caused by medical devices. (References in [1])

Iversen et al. Emerging Themes in Epidemiology 2008 5:22 doi:10.1186/1742-7622-5-22