Background: Numerators and denominators used to estimate infections' incidence rates (IRs), incidence rate ratios (IRRs), and differences (IRDs) vary. Our objective is to quantify the impact of various common definitions for illness episodes (numerators) and person-time at risk (denominators) in estimating these measures.
Methods: Data were from a cohort study in which daily occurrence of illness and children's attendance in day care centers were recorded. We compared 4 IR estimates using various definitions of episode and at-risk time units.
Results: IRs for diarrhea and colds were highest using child-days, lowest for diarrhea using child-weeks, and lowest for colds using child-months. The 4 methods led to similar IRRs but considerably different IRDs.
Conclusion: Incidence rate differences and ratios for infectious diseases can vary by the definition of episode and person-time at risk. This limits the value of the evidence base on which public health recommendations are formulated. Development of a more standard approach to measurement and reporting of IRs is recommended.