Surveillance methods for central venous access device-associated infections in Canadian pediatric hospitals.

The most common complication of central venous access device (CVAD) use is infection, which occurs in 3 to 48% of hospitalized patients. It is recommended that regular surveillance of adverse events with CVADs be conducted, expressed as a proportion of 1000 device days and reviewed and acted upon by the institution's infection control committee. In the process of developing a CVAD program the authors attempted to determine the standard of practice at other Canadian pediatric hospitals. A telephone survey of infection control practitioners (ICPS) or CVAD nurses in 15 university-affiliated Canadian pediatric hospitals was conducted using a standard questionnaire. Fourteen hospitals (93%) conduct surveillance for infections associated with CVADS. One program, a pilot project, follows mechanical complications of CVAD use. Eleven centres conduct comprehensive surveillance; in three, selected patients are followed. Only three programs have sufficient staff to follow out-patients. Definitions for CVAD infections varied widely. A positive blood culture from the catheter is sufficient for diagnosis in eight of the 14 centres (57%); the rest use Centers for Disease Control and Prevention (CDC) or modified CDC criteria. In the four centres where CVAD line days are collected on most or all patients, multiple personnel other than the ICP assist in data collection. Four hospitals report number of infections per 100 discharges, four report absolute number of infections and two use more than one denominator. Surveillance methods rely largely on paper-based chart and microbiology record review; no hospital had access to computerized patient data for direct data retrieval. Eight centres have CVAD committees for policy development, and all 15 have or are developing hospital-wide protocols for CVAD use. Canadian pediatric hospitals recognize the importance of CVAD infections, but it appears that insufficient resources are available to meet recommended data collection methods. Interhospital comparison of rates is not possible at present because of variation in definitions and denominators and in types of patients surveyed.