Expected duration of hospital stay of low birth weight infants: Graphic depiction in relation to birth weight and gestational age

Clinical and laboratory observations Expected duration of hospital stay of low birth weight infants: Graphic depiction in relation to birth weight and gestational age COL James S. Rawlings, MC, USA, COL Franklin R. Smith, MC, USA, and LTC Jose Garc ia , MC, USA From the Newborn Medicine Service, Department of Pediatrics, Madigan Army Medical Cen- ter, Tacoma, Washington, and the Neonatology Service, Department of Pediatrics, Tripler Army Medical Center, Honolulu, Hawaii Neonatal duration of hospital stay correlated with both birth weight and gesta- tional age in positively skewed, nonlinear relationships. Within increments of birth weight, gestational age had a semiindependent influence on length of stay. Log length of stay correlated with both birth weight and gestational age in linear, normally distributed relationships. A nomogram is provided for predicting individual lengths of stay. (J PEDIATR 1993;123:307-9) Despite the high cost of neonatal intensive care, little prac- tical information has been published on the expected length of stay in the hospital among low birth weight infants. In- cremental birth weight-specific mean LOS data have been published. 1-4 Gestational age-specific LOS data and the variance of neonatal LOS have not been reported. The ob- jective of this study was to develop a simple mathematical relationship that could be used to depict graphically, in a practical format, the variance of LOS in relation to both BW and GA. METHODS Data for LOS were prospectively compiled for LBW in- fants delivered during the period from 1981 through 1991 at Madigan Army Medical Center, a regional referral cen- ter with a delivery rate of more than 2500 births per year, a pediatric residency training program and a level I I I neo- natal intensive care unit. The study was limited to surviving inborn N ICU patients with BW 3 0 8 Rawlings, Smith, and Garcia The Journal of Pediatrics August 1993 o �9 120 110 100 90 80 70 60 50 40 30 20 10 0 Z 8 weeks 29 -31 weeks / ~ ~ 3 5 4232-34 weeks weeks I I I I l l l l l l [ IE l l l l l I l l l 2.5 2.0 1.5 1,0 0.5 Birth Weight (Kilograms) Fi 9. I, Mean LOS within increments of GA versus BW for sur- viving, nonanomalous LBW NICU patients. Positive skewing of data and increasing variance in the lower ranges of BW and GA are not depicted. out the ranges of BW and GA and increasing variance in the lower ranges. Figure 1 shows the results of multiple linear regression analyses of incremental GA-specific mean LOS plotted continuously against BW. The discontinuous plots of GA-specific LOS form overlapping linear regressions that emphasize the semiindependent effects of BW and GA on mean LOS. Data comprising overlapping segments of the GA-specific stratifications were significantly different for LOS (p The Journal of Pediatrics Rawlings, Smith, and Garcia 3 0 9 Volume 123, Number 2 ipating and containing health care costs, there is a growing need to look more closely at the expected LOS of neonates who require intensive care. The prediction of individual neonatal LOS is complicated by a number of both predictable and initially unpredictable clinical variables, each of which influences the progress of growth and recovery. The BW and GA are initially known variables that exert strong, predictable influences on LOS. These variables are semiindependent; that is, they are loosely linked within the normally distributed range of BW for any specific GA. Growth-retarded LBW infants, who are more mature than their normally grown counterparts of like BW, usually meet discharge criteria earlier. Thus, in any attempt to predict individual LOS, it is useful to con- sider simultaneously both BW and GA. The variance of LOS is skewed toward longer LOS across the ranges of BW and GA because of the inordinately long LOS for infants who have unusually severe or chronic dis- orders. Logarithmic conversion of LOS data mathemati- cally straightens the nonlinear relationship of BW and GA to LOS and eliminates the skewing of variance. The anal- ysis yields a linear relationship of log LOS to BW and GA with an essentially normally distributed variance at any given BW or GA. Nomograms similar to that presented in Fig. 2 may be used to predict the anticipated range of LOS for individual LBW infants who require intensive care. Infants who have major clinical complications may be expected to have LOS in the upper range of the variance depicted in the nomo- gram. Growth-retarded infants may be expected to have LOS in the lower range of variance specific to their BW. Because of the generally normal distribution of BW for any specific GA, the BW and GA scales in the nomogram will not precisely coincide for most infants. The limits of anticipated LOS for individual infants would fall within the range of values bracketed by these two variables. Thus the nomogram will account for the expected shorter LOS of most growth-retarded infants whose GA falls well to the left of BW on the horizontal scale. The validity of this procedure for predicting LOS is supported by the high predictive value of data from one of our institutions for LOS at the other in- stitution. REFERENCES 1. Poland RL, Bollinger RO, Bedard MP, et al. Analysis of the effects of applying federal diagnosis-related grouping (DRG) guidelines to a population of high-risk newborn infants. Pedi- atrics 1985;76:104-9. 2. Resnick MB, Ariet M, Carter RL, et al. Prospective pricing system for tertiary neonatal intensive care. Pediatrics 1986; 78:820-8. 3. Phibbs CS, Phibbs RH, Pomerance J J, et al. Alternative to di- agnosis-related groups for newborn intensive care. Pediatrics 1986;78:829-36. 4. Commission on Professional and Hospital Activities. Length of stay by diagnosis by operation: United States Pediatric 1983 (pages 214, 252, 258). 5. Ballard JL, Novak KK, Driver M. A simplified score for assessment of fetal maturity of newly born infants. J PEDIATR 1979;95:769-74. 6. Koops BL, Morgan L J, Battaglia FC. Neonatal mortality risk in relation to birth weight and gestational age: update. J PE- DIATR 1982;101:969-77. 7. Berg RB, Salisbury AJ. Discharging infants of low birth weight: reconsideration of current practice. Am J Dis Child 1971;122:414-7. 8. Hermansen MC, Hasan S. Importance of using standardized birth weight increments to report neonatal mortality data. Pe- diatrics 1986;78:144-5. 9. Pomerance J J, Ukrainski CT, Ukra T, et al. Cost of living for infants weighing 1000 grams or less at birth. Pediatrics 1978; 61:908-10.