Gilbert, William M. MD; Nesbitt, Thomas S. MD, MPH; Danielsen, Beate PhD
OBJECTIVE: To determine gestational age– and birth weight–related pregnancy outcomes and resource use associated with prematurity in surviving neonates.
METHODS: A data set linking birth certificates with maternal and newborn hospital discharge records from hospitals in California (from January 1, 1996, to December 31, 1996) was examined for all singleton deliveries by gestational age (weekly, from 25 to 38 weeks) and birth weight (by 250-g increments from 500 to 3000 or more g). Records were examined for respiratory distress syndrome (RDS), use of mechanical ventilation, length of hospital stay in days, and hospital costs.
RESULTS: As expected, RDS, ventilation, length of hospital stay, and costs per case decreased exponentially with increasing gestational age and birth weight. Specifically, neonatal hospital costs averaged $202,700 for a delivery at 25 weeks, decreasing to $2600 for a 36-week newborn and $1100 for a 38-week newborn. Neonatal costs were $224,400 for a newborn at 500–700 g, decreasing to $4300 for a newborn at 2250–2500 g and $1000 for a birth weight greater than 3000 g. For each gestational age group from 25 to 36 weeks, total neonatal costs were similar, despite increasing case numbers with advancing gestational age. Neonatal RDS and need for ventilation were significant at 7.4% and 6.3%, respectively, at 34 weeks' gestation. Significant “excess” costs were found for births between 34 and 37 weeks' gestational age when compared with births at 38 weeks.
CONCLUSION: Prematurity, whether examined by gestational age or birth weight, is associated with significant neonatal hospital costs, all of which decrease exponentially with advancing gestational age. Because total costs for each gestational age group from 25 to 36 weeks were roughly the same ($38,000,000), opportunity for intervention to prevent preterm delivery and decrease costs is potentially available at all preterm gestational ages.
The costs of prematurity remain a significant burden both financially and socially on the US health care system. Cost estimates to prevent or treat these complications run in excess of $2,000,000,000 annually, with the primary morbidity associated with preterm delivery being respiratory distress syndrome (RDS).1 With the development of the neonatal intensive care unit (NICU) more than 25 years ago, there has been a significant decrease in neonatal mortality. Despite these improvements in mortality, there remains significant morbidity, chance of handicap, and expense at the borders of viability.
Therapies to prevent preterm delivery include the use of tocolytic agents to stop preterm labor. The goal is to delay delivery until the risk of neonatal morbidity or mortality is minimized. An additional treatment to improve premature neonatal survival is the maternal administration of steroids to decrease the risk of RDS.2,3 The use of tocolytic agents and maternal steroids after 34 weeks' gestation is not recommended because significant neonatal morbidity and mortality is minimal after this time, and studies do not show benefit.1–4 With the increase in “managed care” and other efforts to control health care costs, health policy planners are trying to obtain accurate cost estimates in an effort to project future health care needs and costs. In addition, true measures of population-based neonatal morbidity, especially for the period of 34–37 weeks, are lacking. In this study, we examined the hospital costs, frequency of complications, length of stay, and other outcomes in singleton deliveries in California by gestational age at delivery and birth weight.
MATERIALS AND METHODS
The California Office of Statewide Health Planning and Development and the University of California at Davis institutional review board gave approval for this study. A statewide database was used, which consists of linked maternal and neonatal–infant hospital discharge records that are matched to vital statistics birth records. This linkage was established for all single live births in acute-care civilian hospitals that reported to the California Office of Statewide Health Planning and Development during the period January 1, 1996, to December 31, 1996. This database did not include births from home deliveries, out-of-state deliveries, and birthing centers not reporting to the Office of Statewide Health Planning and Development, or deliveries in military facilities (total of less than 2% of all deliveries). The method successfully linked 98.9% of maternal and 98.6% of neonatal–infant hospital discharge records with the vital statistics birth record, resulting in an overall linkage of 97.9% of reported singleton deliveries. A database of more than 543,000 deliveries was generated. The International Classification of Diseases, 9th Revision – Clinical Modification and SAS software (SAS Institute, Cary, NC) were used to query the database, which resulted in a specific data set for statistical analysis.
The analysis was restricted to single live births. Infants and mothers were tracked across subsequent transfers until they were sent home, and the analyzed outcomes were tracked across the birth admission and, if applicable, all subsequent transfers at other hospitals. In addition, only infants who survived the first year of life were included in the calculations for length of stay and costs after birth. Hospital charges were only available from the database, so hospital costs were estimated with cost-to-charges ratios.5 The linked database was searched by two methods. The first was to search the database for all deliveries that occurred between gestational ages 25 and 38 weeks, separated into weekly intervals with a minimum birth weight of 250 g (n = 150,462). After the database was examined for gestational age, it was determined that a portion of the deliveries (1.5%) could not have had the reported gestational age, given the known birth weights. An error in recording the true gestational age must have occurred in the majority of these outliers, and they were removed from the database before analysis, as has previously been reported.6 After the above corrections, the gestational age group included a total of 148,248 patients who delivered at between 25 and 38 weeks' gestation. In addition, we excluded infant deaths, which led to a final study population size of 147,224 deliveries. The second method in which the database was analyzed included all deliveries with birth weights between 500 g and 3000 g or more, again eliminating any records with unacceptable gestational age–birth weight combinations and any records pertaining to infant deaths. These patients (n = 458,366) were separated into groups by 250-g increments.
These two groups were then examined independently with respect to multiple demographic, ante-, intra-, and postpartum diagnoses. In addition, hospital costs and length of stay until a home discharge for the newborn were examined. In an attempt to identify hospital costs associated with nonindicated deliveries, estimated costs for births between 34 and 37 weeks' gestation were compared with births at 38 weeks, with cesarean delivery and inductions removed to eliminate those premature births that might have been medically indicated.
Descriptive results for all variables are reported: mean and median length of hospital stay, mean and total costs,5 and percentages otherwise.
The database was examined in two fashions, as described in Materials and Methods. Table 1 displays all premature deliveries by gestational age. Shown are mean and median length of hospital stay, mean and total hospital costs, and percentages of several neonatal morbidities. All morbidities, costs, interventions, and length of stays decreased with increasing gestational age. Total costs for each week of gestation, however, remained fairly constant until 36 weeks' gestation, after which total hospital costs increased owing to the dramatically increased numbers of deliveries in these gestational age groups (Table 1). Table 2 displays neonatal morbidities, length of hospital stay, and hospital costs by birth weight. As with gestational age, there were decreasing morbidities, hospital costs, and interventions with increasing birth weight. Interestingly, total costs for each birth weight group reached a nadir at 1750–2000 g, with the 500–750-g group costing as much as the 2750–3000-g group.
Table 3 shows the additional costs (costs above that of a 38-week-born infant) incurred for infants born at 34–37 weeks' gestation, compared with those for infants born at 38 weeks. In an attempt to identify hospital costs associated with nonindicated deliveries at 34–37 weeks' gestation, labor inductions and cesarean deliveries were removed (Table 3). When these were removed, it was found that $49,906,000 could be saved if deliveries between 34 and 37 weeks could be avoided.
In this large, population-based study, we examined hospital costs, risk of RDS, need for mechanical ventilation, and rate of cesarean delivery in pregnancies that delivered prematurely during a 1-year period in California. Because of the large number of premature deliveries that occurred during this 1-year period, we were able to separate these premature deliveries by week of gestation and by 250-g birth weight increments. This allowed a more detailed and graduated examination of the hospital costs and neonatal morbidities over this period. As expected, neonatal morbidities, hospital costs per case, hospital length of stays, and interventions all decreased with increasing gestational age and birth weight (Tables 1 and 2). Interestingly, we found that even at periods of gestation (34–36 weeks) when major neonatal morbidity is low, there remain significant neonatal hospital costs (34 weeks =$7200 per case, 36 weeks =$2600 per case) and neonatal lengths of stay (34 weeks = 6 days, 36 weeks = 3 days). Intuitively, our data support the concept that efforts to delay premature delivery might be helpful up to 36 weeks' gestation and at estimated fetal weights up to 2250 g. Finally, this information, especially at the later gestational ages and low normal birth weights, might assist health policy planners and health care providers in their attempts at projecting future health care requirements and avenues for cost containment.
In pregnancies complicated by preterm labor, historical recommendations would support the use of tocolytic therapy up to 34 weeks' gestation, based largely on the concept that significant neonatal morbidity and mortality is minimal after this period.1–4 The question behind these recommendations might be stated as, “Why use a potentially harmful treatment modality like intravenous tocolysis when there seems to be no further neonatal benefit after this time?” Our data might suggest that these decisions to withhold therapy might not be wise when one takes into account overall hospital costs. The average neonatal hospital cost for a delivery at 34 weeks was $7200 dollars (Table 1). If delivery was delayed 1 week, there was a 42% decrease in cost, to $4200. Just as significant, a decrease in costs could be seen between the 35th and 36th weeks (38%). Many of the cases with delivery between 34 and 36 weeks were indicated for maternal or fetal reasons, such as preeclampsia or fetal growth restriction. In these cases, prolonging the pregnancy would not be indicated. However, for cases of idiopathic preterm labor, pregnancy prolongation might be indicated, though our current management of pre-term labor is not always effective.
On a case-by-case basis, the very low birth weight neonates were extremely expensive (Tables 1 and 2). When taken in total, however, the overall costs were not much different with each increase in week of gestation until later in the third trimester, when overall numbers of cases dramatically increased. Total hospital costs at 35 weeks' gestation ($41,100,00) were only marginally greater than total cost at 25 weeks' gestation ($39,900,00), despite having more than 50 times the number of patients in the older age-at-delivery group. The fact that the mean length of hospital stay at 35 weeks was 4 days, whereas the median length of hospital stay was 2 days, suggests that certain neonates remained hospitalized much longer than others, and selective intervention might be important to identify and treat that group with prolonged hospitalizations.
Because the vast majority of deliveries occurred after 34 weeks (Table 1), we wanted to examine the hospital costs for infants born in this period compared with those for infants born at term (38 weeks). Table 3 shows the excess costs (costs greater than the cost of a 38-week delivery) per case and total costs. In a crude effort to identify those patients who presented preterm and delivered on their own, we removed those patients with the diagnosis of labor induction or cesarean delivery. It was thought that all of the inductions, and a majority of the cesarean deliveries, would be “indicated deliveries” at this early gestational age. The remaining patients (Table 3) provided a conservative estimate of possibly preventable premature deliveries for which intervention might be of help. If just this “possibly preventable” group of patients had delivered at 38 weeks' gestation, the hospital costs saved would be almost $49.9 million in 1996.
The maternal costs associated with prematurity are presented in Table 1. These costs include any antepartum hospitalizations that might have preceded the delivery hospitalization. The highest maternal cost per week of gestation occurs at 28 weeks, decreasing thereafter. Although the maternal costs are greater at the early gestational ages, they are only marginally greater than costs at 38 weeks' gestation. This is in contrast to the neonatal costs, which are almost 90-fold greater at 28 weeks' gestation compared with 38 weeks' gestation. This would suggest that efforts to prevent or delay delivery would only marginally increase maternal costs while dramatically decreasing neonatal costs.
The incidence of RDS at 34 weeks' gestation was 7.4%, with mechanical ventilation being required 6.3% of the time. By 36 weeks' gestation, the percent of RDS was still 2.3%, with 2.3% requiring ventilation. These percentages, although quite small, are not insignificant, and it is important that prospective parents be made aware of them if they are going to deliver a neonate at these gestational ages.
The individual benefit of each added week of gestation is very clear at the early gestational ages (Table 1). A neonate delivered at 25 weeks' gestation had a length of stay (92 days) that was 18 days longer than that of a neonate delivered at 26 weeks (76 days). Kilpatrick et al7 reported a similar finding when they examined the costs of prematurity in infants delivered between 24 and 26 weeks' gestation in their institution. They found a hospital length of stay of 120 days at 24 weeks, which decreased to 86 days at 25 weeks' gestation. Interestingly, their neonatal length of stay (80 days) for surviving infants at 26 weeks' gestation was similar to ours (76 days). Additionally, St. John et al8 reported a 77-day length of stay for 57 infants who delivered at 26 weeks' gestation. Clearly, our population of patients, when examined by gestational age group, had one measure of outcomes (neonatal length of stay) that matched the two just-mentioned institutionally based studies. We excluded all neonates who died but included those neonates who were transferred to other hospitals and were tracked to a home discharge; thus, we included the vast majority of surviving neonates in California in 1996, which represents a population-based study of survivors.
One criticism of our study is that hospital charges are reported in the database, and hospital costs are estimated as previously described.5 True hospital costs are hard to identify anywhere in the literature and were not available in our database. Another problem we found with the gestational age group was that a number of the patients in the database reported a very early gestational age at delivery; however, when birth weight and other factors associated with prematurity were examined, the gestational age data point had to be wrong. Alexander et al6 had previously come across this problem and had omitted implausible birth weight–gestational age data. In our study we likewise deleted these implausible data points, as described in Materials and Methods. Up to 15% of the newborns in the very early gestational age groups required removal because of this birth weight–gestational age mismatching. Removing these cases gives a more accurate estimation of costs for these groups.
In summary, we present a large, population-based study on the hospital costs and neonatal morbidity associated with prematurity in California in 1996. This information will be important to health policy planners and state health departments as a reference for dealing with prematurity.
© 2003 by The American College of Obstetricians and Gynecologists. Published by Wolters Kluwer Health, Inc. All rights reserved.