Nosocomial infection in a high risk cohort: an illustration of a sampling method

Patients in two acute care university hospitals were entered into the study using the following method: twenty patients were selected each day from all floors/units in each hospital until the hospital communities had been surveyed only once. A total of 762 hospital patients were identified for the study, and the history of each patient was then abstracted from their charts. The data were analyzed as an historical cohort study design. The two hospital samples were compared to show the effects of the nonrandom sampling method using one hospital as a comparison group for the other. Expected associations between nosocomial infection and known risk factors were compared and contrasted between the hospitals using cumulative incidence rates and relative risk ratios. The average length of stay of the group of patients in each hospital was extended from a mean of ten days to a mean of 30 days. The only significant difference between the two hospital cohorts was an age effect. The advantages and disadvantages of this method for selecting a long staying cohort at high risk for nosocomial infection in acute care facilities are detailed.