Monitoring of performance indices on mean length of stay with CUSUM method

Abstract

The CUSUM method was first developed in the settings of industry quality control, in order to monitor processes and raise alarms when process parameters deviates too much from controlled state values. The objective was to assess the feasibility of using the CUSUM method to monitor the activity of hospital wards using the performance index on mean length of stay (IP-DMS) as parameter, illustrated in a dermatology ward of a university hospital. In a retrospective analysis, weekly mean IP-DMS was monitored over a 6-year period (2008-2013) and tested by the CUSUM method; the test was calibrated by empirical analysis of the observed data and by simulations. On average, there were 26 stays by week in the unit, with weekly mean IP-DMS of 1,09 (standard deviation 0.22). The control-state reference value of IP-DMS was 1 and any deviation greater than +0.1 was considered as inacceptable. In 2009, recurrently too high IP-DMS yielded 8 alarms by the CUSUM test. The unit went through organizational changes that lead to a stabilization during 2010. The test raised 2 alarms in the beginning of 2011, related to an episode of hospital-acquired epidemic infections with multiresistant bacteria that had lengthened hospital stay durations and lead to abnormally high IP-DMS. The CUSUM test is a dynamic, precise monitoring tool for medico-economic processes. Implementation of a prospective surveillance could therefore be useful for managers.