Indicators nurses employ in deciding to test for hyperbilirubinemia.

Objective: To identify the indicators nurses employ in deciding to test healthy full-term newborns for total serum bilirubin in the absence of a written protocol.
Design: Secondary analysis of data available on 130 mother-newborn pairs and informal interviews of 30 postpartum unit nurses.
Setting: Two university teaching hospitals.
Participants: All tested newborns and a 33% random sample of remaining newborns from a control group data set created during a previous study and a convenience sample of postpartum nurses from all shifts.
Measurement: Outcome data were obtained from a review of records. Background data were obtained from a review of records and questionnaires. Nurse data were obtained through a modified form of participant observation.
Results: Ninety-one percent of newborns tested for bilirubin were tested unnecessarily. In logistic regression analyses, variables predictive of nurse-driven total serum bilirubin testing were presence of jaundice, odds ratio (OR) = 31.95 (95% confidence interval, 6.71, 152.03), and feeding frequency, OR = 0.28 (0.11, 0.72). Identifying both presence and location of jaundice simultaneously did not significantly predict testing, OR = 1.82 (0.66, 5.04). Fifty-three percent of nurses who were interviewed identified both the presence of jaundice and feeding as indicators to consider for testing.
Conclusion: Newborns are overtested for bilirubin. Indicators used by nurses in deciding to test a healthy newborn for total serum bilirubin are the presence of jaundice and feeding frequency. Nurses who assess feeding frequency are less likely to order bilirubin testing.