Objective. Nosocomial bloodstream infections (NBIs) are associated with serious morbidity and prolonged length of stay (LOS) in very low birth weight (VLBW) infants. However, the marginal costs and excess LOS associated with these infections have never been measured in different birth weight (BW) categories after adjustment for many of the potentially confounding demographic variables, comorbidities, and treatments. The objective of this study was to measure the marginal cost and excess LOS caused by NBIs in surviving VLBW infants in different BW categories.
Methods. This retrospective study examined data previously collected as part of the Neonatal Intensive Care Quality Improvement Collaborative 2000 and the Vermont Oxford Network clinical outcomes database. Univariate analyses and multiple regression were used to examine the effect of NBIs on hospital costs and LOS. Seventeen neonatal intensive care units that participated in the Neonatal Intensive Care Quality Improvement Collaborative 2000 submitted both clinical and financial data on their VLBW infants who were born from January 1, 1998, to December 31, 1999. This study included data from both university and community hospitals.
Results. NBIs occurred in 19.7% of 2809 patients included in this study. NBI was associated with significantly increased treatment costs for infants with BW 751 to 1500 g. The marginal costs of NBIs, as estimated by multiple regression, varied from $5875 for VLBW infants with a BW of 401 to 750 g to $12 480 for those with BW of 751 to 1000 g. LOS was significantly increased in all BW categories. The excess LOS estimated by multiple regression varied from 4 days in VLBW infants with a BW of 1001 to 1251 g to 7 days in those with a BW of 751 to 1000 g.
Conclusions. NBIs are associated with increased hospital treatment costs and LOS but by varying amounts depending on the BW. Preventing a single NBI could reduce the treatment costs of a VLBW infant by at least several thousand dollars. These savings are a greater percentage of the total treatment costs in VLBW infants with BW 1001 to 1500 g than in smaller infants.
- very low birth weight infant
- neonatal intensive care units
- neonatal intensive care unit treatment costs
Nosocomial bloodstream infections (NBIs) commonly occur among very low birth weight (VLBW) infants,1–3 causing significant mortality and morbidity.3–5 These NBIs have also been reported to increase length of stay (LOS) and hospital charges.2,3,6–10 By 1 estimate, NBIs increased LOS by an average of 14 days and total hospital charges by $25 090.6 Reduction of these NBIs has the potential to reduce mortality, morbidity, and hospital treatment costs associated with treating VLBW infants. A recent quality improvement collaborative project, sponsored by the Vermont Oxford Network (VON), focused on reducing NBIs.11 That project found an association between efforts to reduce NBIs, decreasing NBI rates, and median treatment costs.11,12
Efforts to reduce these NBIs also consume resources. Existing published data do not permit an estimate of the per-case cost savings or reduction of LOS that could be used to justify the costs associated with reducing NBIs. Previous studies either have not adjusted estimates for potential confounders such as demographic characteristics, comorbidities, and treatments2,7 or have used charges or LOS as a surrogate for costs.3,6–9 The purpose of this study was to measure the marginal cost, that excess cost attributable only to NBIs, and excess LOS attributable to NBIs in surviving VLBW infants in different birth weight (BW) categories.
The cost and clinical data used for the current study came from 17 hospitals that participated in the Neonatal Intensive Care Quality Improvement Collaborative 2000 (NIC/Q 2000) quality improvement project of the VON. Thirty-four hospitals participated in the project and were asked to submit billing data for 1998 and 1999 as part of the NIC/Q 2000. Eighteen hospitals submitted financial data that were complete and for the entire study period. One of these hospitals was excluded because of inconsistent data quality.
For inclusion in the study, inborn or outborn infants who were admitted to a participating center must have survived to discharge home from the admitting center. We excluded infants with death or transfer to another hospital before discharge home, early-onset sepsis, major congenital malformation, or missing financial or NBI data (Fig 1). The Institutional Review Boards of the University of Vermont and Children's Hospitals and Clinics approved this study.
Cost measures were created by reducing charges to costs using well-established cost conversion methods.12–14 Participating hospitals provided bills with hospital charges at the “uniform bill” (UB) level. Each of the ∼100 UB codes for neonates was mapped to the departmental level as defined on the financial cost reports filed with the Centers for Medicare and Medicaid Services (Health Care Financing Administration form 2552). Using these reports, we generated departmental cost-to-charge ratios specific to each hospital. Charges were then multiplied by the relevant cost-to-charge ratio and summed to arrive at a measure of total cost for each patient. These methods have been shown to produce valid, high-quality estimates of treatment costs for the neonatal intensive care unit (NICU) care of VLBW infants.12–14 All costs were adjusted to 1999 dollars using the consumer price index for health care.15 Total charges for each hospital were also adjusted for geographic variation using the wage index, which is published annually in the Federal Register and is available on the Internet at www.cms.hhs.gov/medicare/ippswage.asp.
Clinical Data Acquisition and Definitions
Participating hospitals submitted clinical information for their VLBW infants to the network using standard VON data forms. Clinical definitions correspond to those published annually in the VON Manual of Operations.16,17 Clinical management of infants including such items as skin preparation before intravenous catheter placement, criteria for removal of central catheters, antibiotic selection, antibiotic levels, follow-up cultures, imaging studies, and other diagnostic and treatment strategies were at the discretion of each institution. Financial data were obtained from hospitals in the NIC/Q 2000 in the form of UB billing data.12,14 Clinical and financial data were merged by the VON staff, removing all hospital and patient identifiers. All cases with financial data matched a corresponding case in the VON database. The investigators received from VON a study data file that contained both clinical and financial data and that did not identify either the participating hospitals or specific patients by name or by their originally assigned numeric identifier.
This study defined NBI as a bloodstream infection occurring after the third postnatal day with coagulase-negative staphylococcus (CONS), other bacterial pathogens, or a fungus. Because the definition of true bacteremia with CONS has varied among institutions, the VON has established specific diagnostic criteria. Patients with CONS in a blood culture were considered to have an NBI only when they had signs of generalized infection (eg, apnea, temperature instability, feeding intolerance, worsening respiratory distress, hemodynamic instability) and were treated with intravenous antibiotics for at least 5 days after a positive culture.16,17 VLBW infants with chronic lung disease (CLD) were defined as those requiring supplemental oxygen at 36 weeks' corrected gestational age (CGA).16,17 In addition, we classified infants who went home requiring supplemental oxygen at 34 to 36 weeks' CGA as requiring supplemental oxygen at 36 weeks and having CLD. We categorized infants who went home without supplemental oxygen at ≤36 weeks' CGA or who remained hospitalized but did not require supplemental oxygen at 36 weeks' CGA as not having CLD. For all other infants, CLD status was considered unknown. Patients were categorized as being small for gestational age (SGA) when they were below the 10th percentile for weight among US births after matching maternal race, gender, multiple birth category, and gestational age, on the basis of the 1993 US Department of Health and Human Services Natality Dataset.18
Marginal costs were defined as treatment costs that were attributable to NBI and that would not have been incurred in the absence of an NBI. These costs are above the usual fixed costs (eg, equipment, administrative personnel, facility costs) and the usual variable costs (eg, nursing care, laboratory tests, medications) that would be required in varying amounts, depending on BW, presence of other medical or surgical complications, etc, of caring for VLBW infants.
To examine comparability of the infants from the 17-hospital sample to the overall network, we identified a comparison group of 29 586 patients from the VON database who were not a part of the study. This group consisted of infants who met the clinical criteria of the study and who were treated during the study period at any of 261 North American centers not in the study. Cost data were not available for these comparison infants.
Statistical Methods and Analysis
Descriptive statistics and preliminary analyses were performed using SAS version 8.1 (SAS Institute, Cary, NC) and SPSS version 10.1 (SPSS, Inc, Chicago, IL) statistical software. Multiple regression models were used to estimate the marginal cost and increased LOS associated with NBI, after adjusting for potentially confounding covariates. The following demographic variables, comorbidities, and interventions were included in the models: BW, SGA, birth location (inborn vs outborn), gender, maternal race (white vs nonwhite), prenatal care, antenatal steroids, multiple birth, 5-minute APGAR score, respiratory distress syndrome (RDS), CLD, necrotizing enterocolitis (NEC), NEC surgery, other surgery, and any ventilation (high-frequency ventilation or conventional ventilation). BW and gestational age were very highly correlated; thus, only BW was used in the regression analyses. Regression analyses used Stata version 7 (Stata Corp, College Station, TX) to adjust the significance levels of the explanatory variables for the lack of independence between observations obtained from multiple infants within the same hospital. The analyses are based on the generalized estimating equations, which provide a practical method to analyze correlated data.19–21 Cost and LOS, which both were positively skewed, were log transformed before analysis. Residuals from the log-transformed models were determined to be reasonably normal, on the basis of probability plots. Because estimated regression coefficients represent additive effects (ie, estimated mean change in dependent measure per unit change in explanatory variable) and the dependent variables were log transformed, the antilog of these coefficients represents proportional changes in the dependent measures in the original scale of measurement. The estimated proportional increases in cost and LOS attributed to NBIs were converted to absolute increases on the basis of the mean of the covariates for infants within each BW category. All means presented for cost and LOS are geometric means, which are the antilog of the log means for each outcome measure. Similarly, adjusted means derived using regression are the antilog of the predicted values. Regression analyses were run separately within defined BW categories after it was determined that the marginal cost of NBIs was BW dependent. Geometric means presented for estimated cost and LOS closely paralleled medians (data not shown). Statistical significance was determined using α = 0.05 for all analyses.
Seventeen study hospitals submitted data for 4483 VLBW infants during the study period. Of these, 2809 met the study criteria (Fig 1). Individual hospitals contributed between 45 and 315 infants each. Among the excluded cases (N = 1674), 771 infants were transferred to another hospital before discharge home, 714 died before discharge, and 3 did not have a final disposition identified. In addition, 94 infants had a major congenital malformation, 44 had early-onset sepsis, 47 lacked complete financial data, and 1 lacked NBI data, all of whom were also excluded from the study. Forty-three (1.5%) of the 2809 study cases lacked complete information for and were excluded from the multiple regression. Table 1 displays the clinical characteristics of the 2809 study patients.
Study hospitals' NBI rates ranged from 5% to 42%. Median total costs varied from $31 869 to $101 905. Median total LOS across all study hospitals varied from 45 to 87 days. In comparison with the group of 261 VON centers (29 586 patients) not in this study, infants from study centers had statistically (but not always clinically) significant differences in several variables. Study infants had a lower BW (1066 vs 1094 g), gestational age (28.4 vs 28.7 weeks), and LOS (59.4 vs 60.9 days). Study infants were also significantly more likely to be born to a white mother (67.6% vs 52.8%) and to have RDS (73.3% vs 66.7%) and CLD (29.8% vs 23.2%). The NBI rates of the 2 groups did not differ significantly (19.7% vs 20.6%; P = .252). The rates of CONS (13.2% vs 13.2%) and other bacterial (8.3% vs 9.2%) and fungal infections (3.0% vs 2.9%) were also similar.
Types and Frequency of NBIs
NBIs occurred in 553 (19.7%) of 2809 study infants (Table 2). CONS infection occurred most commonly and was identified in 372 infants. Other bacterial pathogens were reported in 232 and fungi in 85 infants. Multiple types of NBI were seen in 125 (22.6%) of 553 cases: 73 (13.2%) cases had both CONS and other bacterial pathogens, 27 (4.9%) cases had both CONS and fungi, 14 (2.5%) cases had other bacterial pathogens and fungi, and 11 (2.0%) cases had all 3 types of NBI. The number of NBIs of the same type are not recorded in the VON database.
NBI and Costs
The observed mean costs for all VLBW survivors with NBI exceeded those of uninfected infants by >2-fold ($104 473 vs $49 934; Table 2). Patients with other bacterial pathogens or fungi generated higher treatment costs than those with CONS (Table 2). Infants with NBI had higher mean costs than uninfected infants in bivariate analysis of costs and clinical characteristics. For example, NBI was associated with a 26% increase ($113 890 vs $143 843) in average costs for the lowest BW category and with an 80% increase ($29 628 vs. $53 405) for infants in the highest BW category (Table 3).
Characteristics associated with increased cost were lower BW, lower gestational age, male gender, antenatal steroids, being outborn, Apgar score <5 at 5 minutes, mechanical ventilation, RDS, CLD, NEC or NEC surgery, and other surgery (Table 3). Being SGA was associated with lower costs than those for appropriately grown infants.
Regression Analyses Adjusting Estimated Cost for Covariates
After adjusting for potential confounders and accounting for clustering of infants at participating hospitals, infants in the 3 highest BW categories (751–1500 g) with NBI sustained significantly higher hospital treatment costs than those without NBI (Fig 2). The marginal increase in mean cost specifically attributed to NBI varied from $6276 for those with BW 1251 to 1500 g to $12 480 for those with BW 751 to 1000 g. In percentage terms, these NBIs increased average costs from 15% in infants with BW 751 to 1000 g to 21% for those with BW 1251 to 1500 g (Table 4). The marginal increase for patients with BW 401 to 750 g was $5875, but this increase was not statistically significant (P = .14; Table 4).
Costs of NBI and LOS
Without adjustment for demographic variables, comorbidities, and treatments, NBI was associated with an increased LOS of 13 to 17 days in the 4 BW categories (Table 3). After adjusting for the same covariates used to adjust estimates of cost, the increase in LOS attributable to NBI was 4 to 7 days (Table 4). The differences in LOS for infected and uninfected infants were statistically significant for all BW categories (P < .05). Hospital costs correlated closely with LOS (Pearson product-moment correlation coefficient = 0.90).
This study estimated that NBIs increase hospital treatment costs by $5875 to $12 480 and LOS by 4 to 7 days in all BW categories of surviving VLBW infants (Table 4). These estimates are similar to the data of Baker et al8 for LOS but below those previously published by other investigators.3,6,7,9,10 We hypothesize that this is attributable to 2 factors. First, we used costs, not charges, to estimate marginal cost. For example, Gray et al6 used hospital charges, not costs, to estimate the additional expense associated with NBIs. Rogowski and Harrison13 found that charges overestimated costs in California hospitals by ∼53%. Assuming that costs and charges retained the same ratio in the study of Gray et al, their estimate of charges, $25 090,6 drops to $16 399, a number not too distant from our estimate of the marginal cost of NBIs. This study is the first to use true cost data from multiple hospitals to estimate the marginal cost of NBIs.
The second factor accounting for our lower estimates of marginal cost and excess LOS associated with NBIs is that we included in our multiple regression many potential confounders that were not included in previous studies. We found that NBI increased LOS by 13 to 17 days before controlling for possibly confounding covariates (Table 3). Those numbers were comparable to the 14 to 24 days reported by others who included fewer variables in their analyses.3,6–10 The advantage of including so many covariates is that we have demonstrated that NBI truly increases treatment costs and LOS. NBI is not simply a marker for more seriously ill infants. The disadvantage is that our estimates may be a lower bound of the true increase in cost and LOS associated with NBI.
Our study is the first to delineate the marginal cost of NBIs in different BW categories of VLBW infants. BW has been shown to be a good severity adjustment, even when compared with other, more complicated stratifying variables, such as the SNAP or CRIB scores.22 Although the marginal cost of NBIs was next to the lowest in the largest infants (BW 1251–1500 g), NBIs had the greatest impact on overall costs for these infants (Table 4). Their total treatment costs were much lower than those for infants with lower birth weights; therefore, in percentage terms, the impact of NBIs was greatest in the largest infants (Figs 2 and 3). NBIs increased treatment costs for all infants, but for infants in the BW category 401 to 750 g, the change was a smaller percentage of the total costs and was not statistically significant. The costs for these infants are already so high that the contribution of NBI may be muted by the costs associated with other complications that they experience. Alternatively, because these small infants are more likely to die from NBI, it is possible that they were underrepresented because they died, rather than experiencing an increased treatment cost or LOS.
Our data suggested that NBI increased costs in large part by increasing LOS. NBI was associated with an increase of 4 to 7 days in LOS in all BW categories (Table 4). Because even CONS bacteremias can interfere with enteral feedings and cause apnea and respiratory deterioration, it seems reasonable that CONS and other NBIs could prolong LOS. Although CONS is rarely associated with death,3 it has been shown to increase LOS and hospital treatment charges.2,3,6–10 NBIs in this study had the greatest impact on LOS in the largest infants. The percentage increase in LOS for infants with BW 401 to 750 g was only 6.8% compared with 16.1% for infants with BW 1251 to 1500 g (Table 4).
Our results report an association between NBIs and increased costs and LOS. Our conclusions were based on multiple regression analyses and double checked by propensity scoring analysis (data not shown), which was also consistent with a causal relationship between infection and costs. Although our data do not prove that this is a causal relationship, this seems most likely to be the case. A limitation of our study was that data in the VON database are not timed. It is not possible to know with certainty which events preceded infection and which followed NBI. However, NBIs are not evenly distributed across the duration of the hospitalization. They usually occur early in the hospital course, within the first 3 to 4 weeks after birth.2,3,6,23 Therefore, it seems reasonable that they could increase treatment costs and prolong LOS.
In adults, NBIs increase treatment costs more than other nosocomial infections.24 This may also be true in VLBW, because NBI may cause other complications. NBI may increase the risk of CLD,3,25 patent ductus arteriosus,3,25 and the need for prolonged intravascular access,3 all of which may lengthen LOS and raise treatment costs. Therefore, NBIs may indirectly raise treatment costs by increasing the risk of other complications as well as by directly increasing costs. By adjusting for CLD and other comorbidities in our estimates, we may have underestimated the true influence of NBIs on costs and LOS.
The association of antenatal steroids (Table 3) and increased costs and the association of being SGA and decreased NBIs remains unexplained. However, it is likely that the association of being SGA and decreased NBIs represents an artifact of our stratification by BW. In another study that grouped patients by gestational age, SGA infants acquired nosocomial infection more often than appropriately grown infants of similar gestational age.26
Our study was based on a large, multi-institutional population of VLBW infants from NICUs that were participating in a collaborative project to improve neonatal intensive care, the NIC/Q 2000. The 17 study hospitals represented a self-selected group with serious commitment to quality improvement. Although these hospitals used the same clinical definitions for submitted data, each institution used its own patient management strategies and policies. To account for these and other differences, analyses were adjusted for clustering of patients in the participating hospitals. These hospitals may or may not be representative of other hospitals across the United States, and we found some statistically significant differences between study and nonstudy patients. However, study and nonstudy patients did not differ in their NBI rates.
Study hospitals varied in their size, patient volume, and baseline costs. They also varied in their rates of NBIs as has been reported in other studies.1,3,27 NICUs can show as much as a 10-fold variation in the rate of NBIs. This study's range among participating NICUs, 5% to 42%, was not dissimilar to that reported by previous studies. Despite this variation in hospital rates, analyses adjusted for clustering confirmed that NBI increases costs.
The work of the first NIC/Q project sponsored by VON suggested that intense, multidisciplinary efforts can reduce NBIs11 and NICU treatment costs.12 Similar results were seen with the second NIC/Q collaborative, the NIC/Q 2000.28 Ng et al29 also suggested that NBIs can be reduced. Reducing NBIs should reduce costs and complications associated with treating VLBW infants, thereby improving VLBW outcomes.
Our study was not designed to measure the national costs of NBIs in VLBW infants. However, assuming that the results of this study apply across the United States and using population data from the 1999 Natality Report,30 we estimate that NBIs added almost $100 000 000* to the cost of treating surviving infants with BW 500 to 1499 g, in 1999 dollars.
A 25%11 reduction in these NBIs could result in a savings of at least $24 000 000 (in 1999 dollars) in the United States alone. Because of the skewed nature of the cost data (long right tails), the geometric mean is less than the arithmetic mean. Thus, our estimate is probably a lower bound of the true cost savings possible. The true monetary and emotional costs of NBIs are likely much higher, because NBIs also cause significant mortality.3,4 Such cost savings justify extensive expenditures and efforts to prevent NBIs in VLBW infants.
Reducing NBIs is clinically important, financially prudent, and medically possible. Reducing this unnecessary burden on VLBW infants and their families and society should be a national priority.
This research was funded in part by a grant from the Vermont Oxford Network, a grant from the Children's Hospitals and Clinics to Dr Payne, and a grant from the David and Lucile Packard Foundation to the Vermont Oxford Network.
We also acknowledge the generosity and cooperation of personnel from the 17 institutions whose data were used to conduct this study.
- Received March 7, 2003.
- Accepted January 14, 2004.
- Reprint requests to (N.R.P.) NICU Office, Children's Hospitals and Clinics, 2525 Chicago Ave South, Minneapolis, MN 55404. E-mail:
↵* Calculated as (number of births with BW = 500-999 g from 1999 Natality Report, Table 4323) × survival rate (for infants in that BW category calculated from data submitted by all 290 US hospitals that submitted data to VON in 1998–1999) × infection rate (for infants in that BW category calculated from data submitted by all 290 US hospitals that submitted data to VON in 1998–1999) × mean marginal cost per survivor (geometric mean of the adjusted marginal cost for infants with BW of 500–999 g from the current study sample) + (number of births with BW = 1000–1499 g from 1999 Natality Report, Table 4323) × survival rate (for infants in that BW category calculated from data submitted by all 290 US hospitals that submitted data to VON in 1998–1999) × infection rate (for infants in that BW category calculated from data submitted by all 290 US hospitals that submitted data to VON in 1998–1999) × mean marginal cost per survivor (geometric mean of the adjusted marginal cost for infants with BW of 1000–1499 g from the current study sample). The actual calculation was (22 815 infants × 85.8% survival rate × 37.1% infection rate × $9829 marginal cost) + (28 750 infants × 98.1% survival rate × 11.8% infection rate × $7024 marginal cost) = $94 758 548.
- ↵Gaynes RP, Edwards JR, Jarvis WR, Culver DH, Tolson JS, Martone WJ. Nosocomial infections among neonates in high-risk nurseries in the United States. National Nosocomial Infections Surveillance System. Pediatrics.1996;98 :357– 361
- ↵Stoll BJ, Hansen N, Fanaroff AA, et al. Late-onset sepsis in very low birth weight neonates: the experience of the NICHD Neonatal Research Network. Pediatrics.2002;110 :285– 291
- ↵Goldmann DA, Freeman J, Durbin WA Jr. Nosocomial infection and death in a neonatal intensive care unit. J Infect Dis.1983;147 :635– 641
- ↵Gray JE, Richardson DK, McCormick MC, Goldmann D. Coagulase-negative staphylococcal bacteremia among very low birth weight infants: relation to admission illness severity, resource use, and outcome. Pediatrics.1995;95 :225– 230
- ↵Horbar JD, Rogowski J, Plsek PE, et al. Collaborative quality improvement for neonatal intensive care. NIC/Q Project Investigators of the VON. Pediatrics.2001;107 :14– 22
- ↵Rogowski JA, Horbar JD, Plsek PE, et al. Economic implications of neonatal intensive care unit collaborative quality improvement. Pediatrics.2001;107 :23– 29
- ↵Rogowski J, Harrison E. Treatment Costs for Very Low Birthweight Infants: The California Medicaid Experience. Arlington, VA: The Rand Corporation; 1995 (MR-451-AHCPR)
- ↵Rogowski JA. Measuring the cost of neonatal and perinatal care. Pediatrics.1999;103(suppl) :329– 335
- ↵US Census Bureau. Statistical Abstract of the United States. Washington, DC: US Census Bureau; 2000:112
- ↵Vermont Oxford Network Database Manual of Operations for Infants Born in 1998. Burlington, VT: Vermont Oxford Network; 1997
- ↵Vermont Oxford Network Database Manual of Operations for Infants Born in 1999. Burlington, VT: Vermont Oxford Network; 1998
- ↵1993 US Department of Health and Human Services Natality Dataset. (Available from National Technical Information Service, 5285 Port Royal Rd, Springfield, VA 22161; Phone: 703-605-6000; Product No. PB97-502983)
- ↵Liang KY, Zeger SL. Longitudinal data analysis using generalized linear models. Biometrika.1986;73 :13– 22
- Rogers WH. Regression standard errors in clustered samples. Stata Tech Bull. 13:19–23. Reprinted in Stata Tech Bull Reprints.1993;3 :88– 94
- ↵Pollack MM, Koch MA, Bartel DA, et al. A comparison of neonatal mortality risk prediction models in very low birth weight infants. Pediatrics.2000;105 :1051– 1057
- ↵Lee SK, McMillan DD, Ohlsson A, et al. Variations in practice and outcomes in the Canadian NICU network: 1996–1997. Pediatrics.2000;106 :1070– 1079
- ↵Kilbride HW, Wirtschafter DD, Powers RJ, Sheehan MB. Implementation of evidence-based potentially better practices to decrease nosocomial infections. Pediatrics.2003;111 (4). Available at: pediatrics.org/cgi/content/full/111/4/SE1/e519
- ↵Ventura SJ, Martin JA, Curtin SC, Menacker F, Hamilton BE. Births: final data for 1999. Natl Vital Stat Rep.2001;49 :78
- Copyright © 2004 by the American Academy of Pediatrics