Linking healthcare associated norovirus outbreaks: a molecular epidemiologic method for investigating transmission

Abstract

Background: Noroviruses are highly infectious pathogens that cause gastroenteritis in the community and in semi-closed institutions such as hospitals. During outbreaks, multiple units within a hospital are often affected, and a major question for control programs is: are the affected units part of the same outbreak or are they unrelated transmission events? In practice, investigators often assume a transmission link based on epidemiological observations, rather than a systematic approach to tracing transmission.Here, we present a combined molecular and statistical method for assessing:1) whether observed clusters provide evidence of local transmission and2) the probability that anecdotally|linked outbreaks truly shared a transmission event.

Methods: 76 healthcare associated outbreaks were observed in an active and prospective surveillance scheme of 15 hospitals in the county of Avon, England from April 2002 to March 2003. Viral RNA from 64 out of 76 specimens from distinct outbreaks was amplified by reverse transcription-PCR and was sequenced in the polymerase (ORF 1) and capsid (ORF 2) regions. The genetic diversity, at the nucleotide level, was analysed in relation to the epidemiological patterns.

Results: Two out of four genetic and epidemiological clusters of outbreaks were unlikely to have occurred by chance alone, thus suggesting local transmission. There was anecdotal epidemiological evidence of a transmission link among 5 outbreaks pairs. By combining this epidemiological observation with viral sequence data, the evidence of a link remained convincing in 3 of these pairs. These results are sensitive to prior beliefs of the strength of epidemiological evidence especially when the outbreak strains are common in the background population.

Conclusion: The evidence suggests that transmission between hospitals units does occur. Using the proposed criteria, certain hypothesized transmission links between outbreaks were supported while others were refuted. The combined molecular/epidemiologic approach presented here could be applied to other viral populations and potentially to other pathogens for a more thorough view of transmission.

Figure 1

Gantt display of temporal patterns of gastroenteritis outbreaks. The varying shades of blue background represent hospitals within each NHS trust and each horizontal line represents an inpatient unit. Blue sections are 'outbreak-free periods' and orange sections are 'outbreak periods' (from the 1^st to the last date of onset). A high degree of temporal clustering can be observed in all Trusts. In other words, outbreaks do not often appear in isolation but rather many units are affected sequentially.

Figure 2

Characterised norovirus outbreaks in two hospitals in Avon England April 2002 to March 2003. Each row depicts the follow-up of a single hospital unit. Colored bars represent the period between the onset of illness in the first and last case in an outbreak where norovirus was characterised. Each unique norovirus sequence is represented by a different color. Series of outbreaks meeting the definition of a cluster are circled and were tested for statistical significance.

Figure 3

Sensitivity of the estimate of the probability of a transmission link between outbreaks given the range of prior assumptions of the link.