This article provides an overview of neonatal intensive care unit treatment costs for hospitals that participated in the Neonatal Intensive Care Quality Improvement Collaborative Year 2000 (NIC/Q 2000) quality improvement collaborative and discusses how economic information can be used in quality improvement efforts. Detailed information on neonatal intensive care unit treatment costs is presented for 29 hospitals that participated in the NIC/Q 2000 collaborative. The sample consists of 6797 very low birth weight infants (1500 g or less at birth) with admission dates between January 1, 1997, and December 31, 1998. Information on median treatment cost per infant, ancillary costs, accommodation costs, length of stay, and cost per day is presented. In addition, ancillary costs are further disaggregated into those for respiratory therapy, laboratory, radiology, pharmacy, and all other ancillary services. The role of level of care and other factors that influence treatment costs are then explored.
KEY POINTS OF ARTICLE
Treatment costs for very low birth weight infants are high.
Quality improvement efforts may alter treatment costs in either direction.
APPLYING LESSONS LEARNED TO PRACTICE
Important factors to consider when benchmarking economic performance include mix of patients, birth weight, gestational age, survival, and type of unit.
Advances in neonatal technology in the past decades have significantly improved survival prospects for infants who are born prematurely, but these improvements have come at a high cost. Neonatal intensive care stays are among the most expensive types of hospitalizations.1 Quality improvement efforts will affect treatment costs because they will alter, in some way, the resources used in patient care. A study of a recent collaborative quality improvement effort in neonatology2 demonstrated that significant cost savings may accrue from quality improvement efforts in high-cost patient populations. In this study, each dollar spent on quality improvement in neonatal intensive care yielded $9 in savings in treatment costs during the postintervention year.2
This article discusses treatment costs for neonatal intensive care units (NICUs) that participated in the Neonatal Intensive Care Quality Improvement Collaborative Year 2000 (NIC/Q 2000) collaborative for quality improvement, with a focus on how economic information can be used to inform quality improvement efforts. Information on the magnitude of treatment costs for infants with very low birth weight varies among hospitals in the collaborative. Treatment costs are disaggregated into their major components, including accommodation costs and ancillary costs (respiratory therapy, laboratory, radiology, pharmacy, and all other ancillary services). The article then discusses how information on treatment costs can be used to understand economic performance of a unit and to inform quality improvement efforts.
Thirty-four hospitals participated in the NIC/Q 2000 collaborative quality improvement project. Of these, 32 provided data on treatment costs for the study’s preintervention period, 1997 to 1998. Among the hospitals that submitted data, 3 provided data that were either incomplete or of poor quality and were excluded from these tabulations. Data are presented for 6797 infants with very low birth weight (1500 g and under) from 29 hospitals in the NIC/Q 2000 collaborative.
Hospitals submitted UB-92 billing data for each very low birth weight infant in their Vermont Oxford Network database. The hospital billing data consists of hospital charges. These charges were converted to measures of treatment costs using cost-to-charge ratios from hospital cost reports submitted to the Centers for Medicare and Medicaid Services (CMS). The cost report is a standard financial reporting form generated each year by hospitals as part of reporting requirements imposed by the Centers for Medicare and Medicaid Services (form HCFA-2552). A detailed description of the cost conversion methodology can be found in a recent article on measuring treatment costs for neonatal intensive care.3 The advantage of using measures of treatment cost created in this manner is that they are independent of differences in pricing across hospitals and thus uniform across institutions. All dollar amounts are converted into 1998 constant dollars using the medical component of the consumer price index as generated by the Bureau of Labor Statistics.4 Information on treatment costs is reported using median values instead of mean values because of the long right tails in cost data. Median values are less sensitive to large cost outliers than means and therefore provide a better basis on which to judge economic performance.
TREATMENT COSTS IN THE NIC/Q 2000 COLLABORATIVE
Costs of care for very low birth weight infants among hospitals in the NIC/Q 2000 collaborative are presented in Table 1. Overall, the median treatment cost per infant for infants with birth weight 1500 g and under was $53 300 for hospital care. The median length of stay was 47 days with a median cost per day of $1250. The majority of treatment costs, an average of $37 700, are associated with accommodations. An additional $14 300 in costs is accounted for by ancillary services.
In interpreting information on treatment costs at the unit level, it is important to consider the factors that influence treatment costs. Important among these considerations is the mix of patients in the unit. As shown in Table 1, treatment costs vary widely by birth weight. Thus, the birth weight mix of infants in a unit will affect the average level of treatment costs. Infants with birth weights between 501 and 750 g are the most expensive to care for, with median treatment costs of $103 600. Treatment costs drop as birth weight increases. Thus, whereas infants with birth weight between 751 and 1000 g have median treatment costs of $79 400, for those in the highest birth weight range (1251–1500 g), costs are less than half as much, $31 200. Treatment costs are lowest for infants with birth weight <500 g, largely because these infants have high mortality rates, as reflected in an average length of stay of 2 days. In the sample of infants from the 29 NIC/Q 2000 hospitals, 2% have birth weight <500 g, 19% have birth weight 501 to 750 g, and approximately one quarter have birth weight in each of the higher 250-g birth weight ranges. A unit that has a different birth weight distribution than the average should expect to have different treatment costs. For instance, a unit with a disproportionately high number of infants in the 501- to 750-g range will have higher treatment costs as a result of the mix of patients treated.
Differences in overall treatment costs are attributable to a combination of differences in length of stay and intensity of care, as measured by treatment cost per day. For infants with birth weight >500 g, both length of stay and cost per day drop as birth weight rises. For instance, infants with the highest overall treatment cost per infant (in the 501- to 750-g range) also have the highest median length of stay, 81 days, and treatment cost per day, $1640. Both length of stay and cost per day drop by 20% in the next highest birth weight range (751–1000 g) to 66 days and $1330 per day, respectively. For very low birth weight infants in the highest birth weight range, 1251 to 1500 g, length of stay is again half as high, 32 days. Treatment intensity, as measured by cost per day, is lower but only by 20% ($1050 per day).
Table 2 provides some insights into differences in ancillary costs by birth weight. Median ancillary costs are $14 300. These costs are also inversely related to birth weight for birth weights >500 g. For instance, for infants in the 501- to 750-g range, median costs are $34 400, compared with only $6000 for infants in the 1251- to 1500-g range, a 5-fold difference in magnitude. Respiratory therapy costs differ dramatically by birth weight. The highest median value is for infants with birth weight between 501 and 750 g, $9200. In contrast, respiratory therapy costs for the next highest birth weight range are one third lower ($6100). For infants in the 1001- to 1250-g range, respiratory costs drop by a factor of 3 to $2250 and for the highest birth weight range, 1251 to 1500 g, they are only $840. Pharmacy costs follow a similarly strong pattern with birth weight. For infants with birth weight between 501 and 750 g, median pharmacy costs are $5600. This compares to $4400 for infants in the next 250-g birth weight range and $2800 for those with birth weight between 1251 and 1500 g. Pharmacy costs are likely higher for lower birth weights as a result of the costs of surfactants.
Another measure of patient mix is the gestational age distribution of patients in a unit. As shown in Table 1, treatment costs vary considerably by gestational age. The patterns of treatment costs for gestational age are similar to those for birth weight. Treatment costs rise with gestational ages 24 weeks and above. For infants <24 weeks’ gestational age, treatment cost per infant is the lowest, $8600, as a result of high mortality rates. For these infants, median lengths of stay are only 2 days. However, it should be noted that the treatment intensity is the highest for these infants, with a median daily cost of $1780. Overall treatment costs per infant are highest for those with gestational ages between 24 and 26 weeks, with a median value of $101 600. Both lengths of stay, 79 days, and cost per day, $1560, are high. Treatment cost per infant drops markedly, by 40%, with the next highest gestational age range, 27 to 29 weeks. Infants in that gestational age range have much lower median treatment cost per infant, $63 000. Costs drop again markedly as gestational age rises more, to $34 300 for infants with gestational age of 30 to 32 weeks and $18 700 for those with gestational age >32 weeks. Given the strong relationship between treatment cost and gestational age, the gestational age distribution of infants in a given unit will influence treatment costs. To the extent that a unit disproportionately treats infants with low gestational ages (but above 24 weeks), treatment costs will be higher.
Ancillary costs also vary inversely with gestational age for infants >24 weeks’ gestation. For infants 24 to 26 weeks’ gestational age, ancillary costs are $33 500. For those 27 to 29 weeks, ancillary costs are only half as much, $17 160. This drops again by half for gestational ages between 30 and 32 weeks ($6850) and again by half for gestational ages >32 weeks ($3140). Respiratory therapy and pharmacy costs are particularly high for infants with gestational age between 24 and 26 weeks. The median respiratory therapy cost is $8860 and pharmacy cost is $5590. In contrast, for infants in the 27- to 29-week range, costs are much lower (with a median of $3920 for respiratory care and $3530 for pharmacy). For infants >32 weeks’ gestation, respiratory costs are only $330 and pharmacy costs are $480.
Another factor that may influence costs is the mix of inborn and outborn infants, although the relation is unclear. On the one hand, infants who are outborn may spend somewhat less time in the unit than those who were inborn, but they may be more severely ill and thus more expensive to treat. The data in Table 1 show that for the hospitals in this study, the latter effect dominates the former. Although lengths of stay are lower for outborn infants (43 vs 48 days), treatment intensity is higher. Median costs per day for an outborn infant are $1500 compared with $1200 for inborn infants. The higher treatment intensity offsets the lower length of stay to yield a higher treatment cost per infant for outborn infants ($56 200) versus inborn infants ($52 700). Similarly, infants who are transferred out of a unit should have different treatment costs than those who are not transferred. Table 1 shows that infants who are transferred out have substantially lower lengths of stay, 29 days, and correspondingly lower treatment cost per infant, $41 800. Thus, the mix of inborn and outborn infants and those transferred out of the unit should be taken into consideration when comparing economic performance across units.
In summary, the distributions of birth weight and gestational age and mix of inborn and outborn infants should be taken into consideration when making cost comparisons across units.
ROLE OF MORTALITY
Very low birth weight infants have high mortality rates. In this sample of infants, 12% died during their initial hospitalization. Very low birth weight infants die predominantly in the first days of life. Therefore, those who do not survive their initial hospitalization have much shorter lengths of stay on average than those who do survive. This translates into lower treatment costs. Thus, the mortality status of infants in a unit will influence overall observed treatment costs per infant at the unit level.
Table 1 provides some insights into this relationship. Three percent of infants with birth weight <1500 g died during their first day of life. Treatment intensity was high for these infants, a median of $2000 for the 1-day hospital stay. Another 9% of infants died after the first day of life. The median length of stay for these infants was only 7 days. Although the cost of the hospitalization was relatively low, $18 000, treatment intensity was very high, with a median value of $2600 per day. This is more than twice as high as the daily cost for infants who lived and were discharged from the hospital ($1100). Because the median length of stay is higher for infants who are discharged from the hospital (57 days), overall treatment costs per infant are higher ($63 800) for infants who live than for those who die.
Table 3 provides additional insights into the role of mortality on treatment costs by considering how costs differ by survival status within birth weight and gestational age ranges. From Table 3, it is evident that there are enormous differences in the costs of survivors and nonsurvivors for infants with the lowest birth weights and gestational ages. For infants with birth weight <500 g, those who survived had costs per infant of $147 660. As shown in Table 4, respiratory and pharmacy costs were particularly high for these infants, with median values of $16 250 and $8980, respectively. Infants who died after the first day of life had median treatment costs of $12 970. Cost per infant for survivors in the 501- to 750-g range were $125 900, compared with $19 000 for those who survived the first day of life. The cost per survivor drops to $83 500 for infants in the 751- to 1000-g range and $51 300 for those in the 1001- to 1250-g range, and $31 900 for those in the highest range (1251–1500 g). The cost per infant who died after the first day of life is roughly comparable across all birth weight ranges. Therefore, units that have better survival rates will have higher treatment costs, particularly if they are disproportionately successful at treating infants with the lowest birth weights.
The same pattern of costs for survivors is evident within gestational age categories. Infants with gestational age <24 weeks and survived had median treatment costs of $180 000. For infants who survive, respiratory therapy and pharmacy costs are particularly high, $17 700 and $10 100, respectively (as shown in Table 4). Infants with a gestational age of <24 weeks and die after the first day of life have overall treatment costs of only $9400. Median treatment costs for infants who survive drop sharply with gestational age. For infants in the 24- to 26-week range, treatment cost per infant is $117 400. This compares to $65 100 for infants in the 27- to 29-week range and $34 600 for those with gestational age between 30 and 32 weeks. Infants with gestational age >32 weeks have median treatment costs of only $19 400. Those who are between 24 and 32 weeks’ gestation and die after the first day of life have median treatment costs between $20 400 and $25 400. This drops to $7100 for those with >32 weeks’ gestation. Thus, unlike birth weight, as gestational age rises, the cost difference between survivors and nonsurvivors gets smaller. This suggests that units with the highest survival rates among infants of the lowest gestational ages will have the highest costs. Improved mortality performance at higher gestational ages will have less of an impact on overall unit costs.
These results suggest that the mortality status of infants, particularly among the smallest infants, will influence overall observed treatment costs per infant at the unit level. From a quality improvement perspective, these results suggest that if quality improvement results in lower mortality rates for the smallest and least mature infants, then treatment costs at the unit level may rise. However, quality improvement also has other economic benefits. To the extent that improvements in patient quality of care increase the efficiency of care, treatment costs will go down. For instance, in a collaborative quality improvement in neonatology, it was demonstrated that overall treatment costs decreased as a result of the quality improvement initiative.2
ROLE OF LEVEL OF CARE
NICUs also vary in the level of care provided. This influences observed treatment costs per infant at the unit level. Table 5 presents information on median treatment costs for infants by level of care. Among the 29 hospitals, there were 13 level C units, 15 level B units, and 1 level A unit. Units with levels A and B are combined in Table 5. Level of care is ascertained yearly by the Vermont Oxford Network from its members in an annual survey. Level of care is self-reported according to the following definitions. Level A units are defined to be those with restrictions on the capability to provide assisted ventilation and that perform only minor surgery. Level B units, in contrast, perform all major surgery for newborns except cardiac surgery requiring bypass. Level C units are capable of providing cardiac surgery requiring bypass for newborns.
As shown in Table 5, treatment costs are higher for level C units than lower level units. Median treatment cost per infant for infants in level C units was $59 100 compared with $47 000 for infants in level A/B units; this represents a 20% difference. The cost differences between unit types are largely driven by differences in the intensity of care. Lengths of stay are generally similar between unit types, with a median of 48 days in level C units and 46 days in lower level units. However, the median cost per day is higher in level C units, a median of $1350 compared with $1160. Accommodation costs are higher in level C units, with a median value of $42 000, compared with $33 000 in lower level units. This represents a difference of 20% and is largely driven by differences in the cost per day, as lengths of stay do not differ widely between unit types. Ancillary costs are higher at level C units by approximately 20% as well ($15 800 compared with $12 700). As shown in Table 6, pharmacy and laboratory costs differ widely between unit types. For instance, median laboratory costs for level C units are $3200 compared with $2250 for level A/B units. Pharmacy costs are $3740 compared with $2130. In contrast, differences in the cost of respiratory care are not as large, with a median value of $2840 for level C units and $2630 for lower level units.
It is worth noting that the birth weight and gestational age distributions of infants treated in the level C and level A/B units are generally similar, as are the mortality distributions. Although we have seen above that birth weight, gestational age, and survival influence treatment costs, differences in patient mix in these dimensions are not driving observed differences in treatment costs. Infants in level C units are somewhat more likely to be outborn (22% vs 18%) and to be transferred out of the unit (21% vs 16%), which may drive some of the observed differences but not all. Thus, differences between unit types are likely to reflect unmeasured factors, such as the severity of patient mix and costs associated with specialized types of surgery such as for cardiac care. Thus, in comparing economic performance across units, it is also important to consider the type of unit in addition to differences in birth weight, gestational age, and mortality differences.
Neonatal intensive care costs for infants with very low birth weight are high. On average, median treatment costs per infant are $53 300 in 1998 constant dollars. When comparing economic performance across units, it is important to take into consideration factors that influence treatment costs. These include birth weight, gestational age, and survival. Treatment costs are inversely related to birth weight for infants with birth weight >500 g. Similarly, treatment costs fall with gestational age for infants with gestational age >24 weeks. Thus, the birth weight and gestational age distributions of patients in a NICU will affect observed treatment costs. To the extent that these factors change over time, overall treatment costs per infant will also change. In some cases, an unusual spike in the low birth weight or earlier gestational age distribution in a given year, particularly for a small unit, could generate unusually high costs for that year. In interpreting time trends, it is therefore important also to consider the trend in the composition of the patients in the unit.
Treatment costs also depend on survival. Because most very low birth weight infants die in the first days of life, overall treatment costs per infant are lower for infants who die than for those who survive. Thus, units with better survival will also have higher treatment costs. Quality improvement efforts that improve survival may therefore increase costs, particularly if survival is improved for infants with birth weight <750 g and gestational age <24 weeks. The difference in median treatment cost per infant for a 501- to 750-g infant who survives versus one who dies after the first day of life is $107 000. The difference in median treatment cost per infant for a infant who has a gestational age between 24 and 26 weeks and survives compared with dying after the first day of life is $170 600. Quality improvement efforts, however, may also increase the efficiency of care, which may counterbalance any increases in cost associated with improved survival. Recent evidence from a collaborative quality improvement in neonatology demonstrated that this is possible.2
In addition, level C units have costs that are higher (by approximately 20%) than level A/B units despite having similar distributions of birth weight, gestational age, and survival. Costs are higher as a result of the intensity of care provided, not lengths of stay. Both accommodation and ancillary costs are higher for these units. Differences between unit types likely reflect unmeasured differences in severity and in the costs associated with specialized types of surgery, such as those for cardiac care.
Information on treatment costs provides a tool to ascertain the economic performance of NICUs and to determine the effects of quality improvement efforts. However, it is important that measures of cost be created in a uniform manner across institutions so that these costs are comparable. In addition, when comparing costs across units, it is vital to take into consideration factors that influence costs, including the distributions of birth weight, gestational age, survival, and the level of care of the unit.
- ↵Office of Technology Assessment. Neonatal intensive care for low birth weight infants: costs and effectiveness. Washington, DC: Congress of the United States; 1987
- ↵Rogowski J, Horbar J, Plsek P, et al. Economic implications of neonatal intensive care unit collaborative quality improvement. Pediatrics.2001;107 :23– 29
- ↵Rogowski J. Measuring the cost of neonatal and perinatal care. Pediatrics.1999;103(suppl) :329– 335
- ↵US Bureau of Labor Statistics. Monthly Labor Review. Washington, DC: US Bureau of Labor Statistics; 1998
- Copyright © 2003 by the American Academy of Pediatrics