To determine factors associated with variation in CHGS of infants with RDS, the medical records of 210 infants were examined. Statistical analysis included Chi-square, multiple linear regression (MLR) and discriminant analysis (DA). The dependent variable (MLR and DA) was the natural log of CHGS per length of stay (LOS). A split-half validation analysis was also performed. Data included 61 patient-related variables and 7 physician-related variables. Chi-square identified 38 patient variables significantly associated with CHGS. In sum, small, sick babies have higher CHGS (p<0.001); however, CHGS also increase with poor prenatal care (p<0.02) and maternal use of drugs or alcohol (p<0.044). Therapeutic interventions for RDS, nutritional support variables and therapeutic complications were all statistically significant(p<0.001). Severity of illness scores (SNAP, NTISS, CRIB) were also significant (p<0.001). Maternal antenatal steroid use was not significantly related to CHGS. However, only 11% of mothers received prenatal steroids. No physician variables were significantly related to CHGS. Stepwise MLR identified 10 significant variables that explained 78.6% of the variance(s2) in CHGS per day (R=0.887, p<0.001). Step-one, severity of illness (SNAP score), explained 35.7% of s2. The MLR also included birth weight, discharge to home, doses of surfactant, discharge weight, days of oxygen therapy, prenatal care visits, peak bilirubin, years of physician experience, and an interaction of an abnormal potassium and PVL. In the validation sample, the MLR explained 77.6% of the s2 (R=0.881, p<0.001). Stepwise DA identified 8 variables that explained 77% of s2 (p<0.001), and correctly classified 96.3% of“grouped” cases (95.1% in validation sample). The DA step-one, severity of illness (NTISS Score), explained 24.2% of s2. The DA also included doses of surfactant, birth and discharge weights, ventilation days, SNAP score, birth weight >1500g and multiple gestation. Conclusion: Patient factors (especially severity of illness, change in weight from birth to discharge, surfactant dosing, and ventilation history) explain most variation in CHGS of premature infants with RDS. Physician variables, except years of experience, are not significant factors.
