Physician variations and the ancillary costs of neonatal intensive care.

OBJECTIVE: To determine to what degree attending physicians contribute to cost variations in the care of ventilator-dependent newborns. DATA SOURCES: Clinical data were merged with hospital financial data describing daily ancillary care costs during the first two weeks of life for 132 extremely low-birthweight newborns. In addition, each patient's chart was reviewed and illness severity graded using both SNAP and CRIB scores. STUDY DESIGN: This was a retrospective cohort of infants with birth weights of less than 1,001 grams and respiratory distress syndrome requiring mechanical ventilation in the first day of life. From birth up to two weeks of life, each received care directed by only one of 11 faculty neonatologists in a single university hospital. Data were analyzed stratified by these physicians. t-Test, ANOVA, and chi-square were used to assess bivariate data. For continuous data, log linear regressions were used. PRINCIPAL FINDINGS: After controlling for illness severity, when stratified by physicians, there were significant variances in the costs of ancillary resources for the study infants (p < .0001). Twenty-nine percent of the variance was attributable to whether or not the hospital day included the use of a ventilator. Physician identity explained only 5.6 percent (p < .0001). CONCLUSIONS: Physician identity was significant but explained less than 6 percent of the total variance in ancillary costs. Whether or not a ventilator was used during care was far more important. We conclude that for very sick babies during the first two weeks of care, reducing variations in ancillary services utilization among neonatologists will yield only modest savings.