Statistical epidemic modeling with hospital outbreak data

Stat Med. 2008 Dec 30;27(30):6522-31. doi: 10.1002/sim.3419.


The analysis of epidemic data has one special feature: individuals are highly dependent, i.e. infected cases are the cause of further infected cases (cross-infection). The main epidemiological parameter of interest is the transmission rate: the rate with which an infectious individual has close contacts with other patients in the hospital unit resulting in colonization or infection. In order to estimate this parameter, the statistical analysis should be based on an appropriate compartmental model that describes the transmission dynamics of an epidemic process. Nonparametric methodology is available for closed populations without migration, but especially in hospitals, admission and discharge have to be taken into account in addition. Transmission and discharge have to be considered as competing events. Martingale-based methodology takes the time-dependent feature of the rates adequately into account and yields useful estimates. These methods are applied to an outbreak of the specific hospital pathogen vancomycin-resistant enterococci (VRE) in an onco-haematological unit at the University Medical Center Freiburg in Germany.

Publication types

  • Research Support, Non-U.S. Gov't

MeSH terms

  • Cross Infection / epidemiology*
  • Cross Infection / transmission
  • Disease Outbreaks / statistics & numerical data*
  • Germany / epidemiology
  • Gram-Positive Bacterial Infections / epidemiology*
  • Gram-Positive Bacterial Infections / transmission
  • Humans
  • Models, Statistical*
  • Vancomycin Resistance*