Original Article

Journal of Perinatology (2013) 33, 514–519; doi:10.1038/jp.2013.2; published online 31 January 2013

Impact of early human milk on sepsis and health-care costs in very low birth weight infants

A L Patel1,2, T J Johnson2,3, J L Engstrom2,4, L F Fogg2, B J Jegier2, H R Bigger1 and P P Meier1,2

  1. 1Department of Pediatrics, Rush University Medical Center, Chicago, IL, USA
  2. 2College of Nursing, Rush University Medical Center, Chicago, IL, USA
  3. 3Department of Health Systems Management, Rush University Medical Center, Chicago, IL, USA
  4. 4Frontier Nursing University, Hyden, KY, USA

Correspondence: Dr AL Patel, Department of Pediatrics, Rush University Medical Center, 1653 W. Congress Parkway, Murdock 622, Chicago, IL 60612, USA. E-mail: Aloka_Patel@Rush.edu

Received 21 August 2012; Revised 21 December 2012; Accepted 28 December 2012
Advance online publication 31 January 2013





To study the incidence of sepsis and neonatal intensive care unit (NICU) costs as a function of the human milk (HM) dose received during the first 28 days post birth for very low birth weight (VLBW) infants.

Study design:


Prospective cohort study of 175 VLBW infants. The average daily dose of HM (ADDHM) was calculated from daily nutritional data for the first 28 days post birth (ADDHM-Days 1–28). Other covariates associated with sepsis were used to create a propensity score, combining multiple risk factors into a single metric.



The mean gestational age and birth weight were 28.1±2.4 weeks and 1087±252g, respectively. The mean ADDHM-Days 1–28 was 54±39mlkg−1day−1 (range 0–135). Binary logistic regression analysis controlling for propensity score revealed that increasing ADDHM-Days 1–28 was associated with lower odds of sepsis (odds ratio 0.981, 95% confidence interval 0.967–0.995, P=0.008). Increasing ADDHM-Days 1–28 was associated with significantly lower NICU costs.



A dose–response relationship was demonstrated between ADDHM-Days 1–28 and a reduction in the odds of sepsis and associated NICU costs after controlling for propensity score. For every HM dose increase of 10mlkg−1day−1, the odds of sepsis decreased by 19%. NICU costs were lowest in the VLBW infants who received the highest ADDHM-Days 1–28.


premature infant; breast milk; infection; economics; health-care costs



Late-onset sepsis (sepsis) occurs in approximately 22% of very low birth weight (VLBW; birth weight (BW) <1500g) infants in the United States, making it one of the most common morbidities in this population.1 In addition to predisposing the infant to other morbidities,2 and subsequent neurodevelopmental disability,3 sepsis significantly increases neonatal intensive care unit (NICU) costs2, 4 and translates into higher societal and educational costs for VLBW infants who survive sepsis with neurodevelopmental disability.5 Thus, interventions to decrease the risk of sepsis are a high priority for VLBW infants.

Several small studies have demonstrated that human milk (HM) feedings reduce the incidence and/or the risk of developing sepsis in extremely preterm and VLBW infants.6, 7, 8 Furthermore, studies and research reviews7, 9, 10, 11 suggest that the early post-birth period may represent a ‘critical period’ for the feeding of high doses of HM to reduce the risk of sepsis and other morbidities in premature infants. This critical period has been variably defined as the first 14 or 28 days of life (DOL). We selected the first 28 DOL as most VLBW infants are receiving full enteral nutrition by that time, as opposed to the first 14 DOL when 55% of our infants were still receiving parenteral nutrition (PN). Furthermore, the cost of treating sepsis is high, due to increased ventilation use and longer lengths of NICU stay. Recent research has demonstrated that sepsis is associated with $10055 in additional costs to the hospital.2 The purpose of this study was to determine the relationships among sepsis, hospital costs and the dose of HM feedings during the NICU hospitalization in a prospective cohort of VLBW infants.




This is an ongoing prospective cohort study of VLBW infants admitted to Rush University Medical Center NICU since February 2008 who survived to receive enteral feedings by DOL 14.12 Exclusion criteria included the following: BW >1500g, birth gestational age (GA) >35 weeks, initiation of enteral feedings after DOL 14, onset of sepsis before initiation of feedings, major congenital anomalies or chromosomal disorders, maternal conditions that precluded HM provision or use for the infant (for example, maternal cocaine use) and death before NICU discharge, because the costs of care for these infants would likely be lower and not comparable for these subjects due to early death.13, 14 Of the three subjects excluded due to death before NICU discharge, none had sepsis. Subjects were excluded after enrollment if they were transferred to a lower-level NICU because complete cost data for the hospital stay would not be available. This study was approved by the Rush University Institutional Review Board. Signed informed consent was obtained from parents/guardians of all enrolled subjects.

Nutritional practices for VLBW infants

In 2008, all VLBW infants received PN upon NICU admission, and mothers were strongly encouraged to provide HM for their infants. Colostrum was administered oropharyngeally once available, even if the infant was not receiving enteral feedings.15 When the neonatologist deemed an infant stable (for example, no longer receiving inotropic support or indomethacin), enteral feedings were started at 20mlkg−1day−1 and advanced daily by 20mlkg−1 as tolerated and PN was decreased in a complementary manner. Except for the 14 subjects who received donor milk products (donor milk and HM-based fortifier) through a separate ongoing clinical trial in 2008,16 feedings were initiated in all the other infants with unfortified HM or 20cal preterm formula. HM was fortified with bovine HM fortifier when feeding volumes reached 100mlkg−1day−1. Formula-fed infants were switched from 20 to 24cal preterm formula when feeding volumes reached 140mlkg−1day−1. Thereafter, the caloric content and feeding volumes were adjusted based on the infant’s growth and tolerance. Hindmilk was given for a maximum of 14 days to a small subset of infants (6/175=3%) for slow growth with exclusive HM feedings at the neonatologist’s discretion.17 Infants received fresh HM if it was available; otherwise they received previously frozen HM. Starting in December 2009, a BW-based enteral feeding protocol, which did not alter the PN or oropharyngeal colostrum provisions, was instituted for all VLBW infants. This protocol consisted of a minimal enteral feeding period (3–4 days), during which time feedings were not advanced, and then a BW-based advancement period, which lasted 9–17 days. HM was fortified with a bovine HM fortifier when feeding volumes reached 140mlkg−1day−1. Formula-fed infants were switched from 20 to 24cal preterm formula when feeding volumes reached 140mlkg−1day−1. Infants were discharged once they achieved satisfactory weight gain with full oral feedings, thermal stability and resolution of acute medical conditions.

Data collection and clinical outcomes

Prospectively collected data included daily weight, occurrence of sepsis, maternal and neonatal clinical factors, daily intake (ml) of clear intravenous fluids, PN, HM and formula, and timing of HM fortifier supplementation. Sepsis was defined as a positive blood culture after DOL 3 and with antibiotic treatment greater than or equal to5 days. Daily dose of HM was calculated as follows: the total number of ml of HM (fortified and unfortified) received by the infant during each 24-h period from 0000 to 2400 hours was summed. This summed value was divided by the infant’s measured weight for that day and expressed as mlkg−1day−1. These individual daily measures were then summed and divided by the number of days to create the average daily dose of HM (ADDHM) for the first 28 days post birth or until discharge from the NICU if the infant was discharged before DOL 28 (ADDHM-Days1–28). Although it would be ideal to calculate the HM dose received by each infant before onset of sepsis, the resulting variation in the study interval would make comparisons among sepsis and non-sepsis infants difficult to interpret. The first 28 days post birth has been used by previous investigators7 in a prospective observational study of VLBW infants and was selected as the exposure period for HM dose calculations for this study. ADDHM-Days 1–28 was subdivided into three dose categories for analysis based on previous literature (<25, 25–49.99 and greater than or equal to50mlkg−1day−1).6, 7 Daily dose of HM was also calculated as the proportion of total enteral intake over a 24-h period that consisted of HM (fortified and unfortified) and compared with ADDHM.

Economic data

Rush University Medical Center’s system-wide cost accounting system provided the direct cost of care for each chargeable item (for example, room and board, all clinical and non-clinical personnel time with the exception of physician’s time) used during each infant’s hospital stay. The direct costs for each chargeable item were summed to calculate the hospital direct cost for each NICU hospitalization. Direct costs included only costs incurred by the hospital, because the physicians used a separate billing system. Direct costs were adjusted to 2010 dollars using the Bureau of Labor Statistics Consumer Price Index for all items.18

Data analysis

Descriptive statistics included mean±s.d. (minimum–maximum), median (minimum–maximum) and number (percent). Categorical data were analyzed using chi square or Fischer’s exact test as appropriate, ordinal data were analyzed using the Mann–Whitney U test and continuous data were analyzed using t-test or analysis of variance to assess differences between subjects with sepsis and subjects without sepsis. Variables with data that were not normally distributed were logarithmically transformed before analysis to produce normal distributions.

Owing to the fact that there were a relatively small number of events and a relatively large number of potential covariates, a propensity score19 was created to account for variables that were statistically and clinically associated with sepsis. Statistically, the propensity score allowed the assignment of a proxy value for the likelihood that each infant in the data set would develop sepsis. Pearson’s correlation coefficients were used to assess the correlation of potential covariates with sepsis. Initially, we examined nine risk factors (GA, BW, gender, race, receipt of surfactant, steroid post birth, DOL of enteral feeding initiation, receipt of PN on DOL 10 and mechanical ventilation on DOL 10) that were potentially associated with sepsis either due to correlation at P<0.1 or due to clinical plausibility. DOL 10 preceded the onset of sepsis for all infants, thus avoiding the potential effect of reverse causality in which the occurrence of sepsis would prolong PN or mechanical ventilation duration. Multiple binary logistic regression analysis was used to obtain the optimal combination of risk factors, which resulted in retention of five risk factors in the final propensity score: BW,20 receipt of surfactant, receipt of PN on DOL 10,21 white race and DOL of enteral feeding initiation.20 Collinearity between these five variables was <7% of the variance in the predictor variance–covariance matrix. ADDHM-Days 1–28 was excluded from the propensity analysis because this was our primary predictor variable of interest.

For the current study, the outcome of sepsis was considered as both a dichotomous and a time-to-event outcome. Multiple binary logistic regression analysis, controlling for the propensity score, was used to evaluate the effect of ADDHM-Days 1–28 on the occurrence of sepsis. The results of the logistic regression analysis were cross-validated with a Cox regression analysis using ADDHM-Days 1–28 and propensity score as predictors. The dependent variable in the Cox regression was the hazard of sepsis.

A generalized linear regression model was fit to evaluate the relationship between direct cost and ADDHM-Days 1–28 (<25, 25–49.99 and greater than or equal to50mlkg−1day−1), controlling for the propensity score and the presence of sepsis. The generalized linear model was fit with a gamma distribution and log link function. The distribution was selected using a modified Park test, which identifies the appropriate distribution for the mean–variance relationship.18, 22 We re-estimated the regression model excluding sepsis to test whether its omission affected the relationship between ADDHM and costs. Average predicted costs were computed for the three ADDHM categories. One-way analysis of variance was used to test for differences in actual costs and predicted costs from the regression models across the three categories, and Tukey and Bonferroni’s methods for post-hoc comparisons were conducted. Analyses were performed using SPSS 15.0 (Chicago, IL, USA) and SAS Version 9.2 (Cary, NC, USA). Statistical significance was set at P<0.05 for all analyses.




During 2008–2010, 285 VLBW infants were enrolled (Figure 1) in this study; 175 infants for whom data collection was complete comprise the study cohort. Demographic, maternal and neonatal characteristics for the study sample and each subgroup (with and without sepsis) are reported in Table 1. Of the 175 infants, 171 (97.7%) received at least some HM, and the ADDHM-Days 1–28 for the entire cohort was 53.6±39.2mlkg−1day−1 (mean±s.d.; min–max 0–135mlkg−1day−1). Figure 2 depicts the average daily HM intake, in terms of both the ADDHM (mlkg−1day−1) and the daily dose of HM as a proportion of total enteral intake, for each ADDHM-Days 1–28 dose category. After the first 5 DOL, infants in the highest HM dose group consistently received more HM on a day-by-day basis than infants in the middle group who received more than the lowest HM dose group. By contrast, infants in all the three HM dose groups received at least 50% of enteral intake as HM for the first 10 DOL, with a subsequent gradual decline during the remaining 18 days for the lower two HM dose groups.

Figure 1.
Figure 1 - Unfortunately we are unable to provide accessible alternative text for this. If you require assistance to access this image, please contact help@nature.com or the author

Flow diagram of subject recruitment. DOL, day of life; GA, gestational age; NICU, neonatal intensive care unit; SGA, small for gestational age; VLBW, very low birth weight.

Full figure and legend (100K)

Figure 2.
Figure 2 - Unfortunately we are unable to provide accessible alternative text for this. If you require assistance to access this image, please contact help@nature.com or the author

Daily mean human milk (HM) intake in three different ADDHM-Days 1–28 (average daily dose of human milk for the first 28 days post birth) dosing groups: filled circle <25mlkg−1day−1, n=57; 25–49.99mlkg−1day−1, n=32; greater than or equal to50mlkg−1day−1, n=86. (a) HM dose expressed as ADDHM (mlkg−1day−1). (b) HM dose expressed as a proportion of total enteral intake.

Full figure and legend (74K)

Clinical outcomes

The occurrence of sepsis and other morbidities, survival, growth and duration of NICU hospitalization are shown in Table 1. Sepsis occurred in 23 (13%) infants at a mean postnatal age of 30±18 days (median 24 days, min–max 12–77). The organisms isolated from sepsis cases were coagulase-negative staphylococci (17%), Staphylococcus aureus (17%), other Gram-positive bacteria, including group B Streptococcus and Enterococcus species (9%), Escherichia coli (17%), Klebsiella species (13%), Enterobacter species (9%), Pseudomonas species (9%) and Serratia species (9%).

After correcting for propensity score, results of the binary logistic regression analysis demonstrated an independent protective effect of ADDHM-Days 1–28 on sepsis, with a 19% reduction in the odds of developing sepsis for every 10mlkg−1 day−1 increase in ADDHM-Days 1–28 (odds ratio 0.981, 95% confidence interval 0.967–0.995, P=0.008) (Table 2). Cox regression analysis correcting for propensity score revealed the same protective effect of ADDHM-Days 1–28 on sepsis with a significant reduction in the hazard ratio (0.982, 95% confidence interval 0.968–0.995, P=0.009). Figure 3 demonstrates this dose–response effect of ADDHM-Days 1–28 on sepsis after correcting for propensity score using ADDHM dose categories based on previous literature.6, 7

Figure 3.
Figure 3 - Unfortunately we are unable to provide accessible alternative text for this. If you require assistance to access this image, please contact help@nature.com or the author

Adjusted survival curves for sepsis by ADDHM-Days 1–28 (average daily dose of human milk for the first 28 days post birth) over the first 28 days of life. Survival curves adjusted for propensity score.

Full figure and legend (13K)

Economic impact

Table 3 reports the mean costs by ADDHM-Days 1–28. Correcting for propensity score and sepsis did not significantly alter the adjusted costs. However, after correcting for propensity score and sepsis, average costs were $31514 lower for infants with ADDHM-Days 1–28 greater than or equal to50mlkg−1day−1 and $20384 lower for infants with ADDHM-Days 1–28 25–49.99mlkg−1day−1, when compared with infants with ADDHM-Days 1–28 <25mlkg−1day−1 (P<0.001). Tukey and Bonferroni’s methods for post-hoc comparisons indicated significant differences in costs for infants with ADDHM-Days 1–28 <25mlkg−1day−1 compared with infants with 25–49.99 and greater than or equal to50mlkg−1day−1. The comparison between 25–49.99 and greater than or equal to50mlkg−1day−1 was not significant.



In this study, we report a dose–response relationship between the amount of HM feedings received during the first 28 days post birth and a reduction in sepsis risk and mean hospital costs for a prospective cohort of VLBW infants. Even after adjusting for several covariates that predicted risk of sepsis, Cox regression analysis revealed that ADDHM-Days 1–28 was a stronger predictor of sepsis than was the combination of variables in the propensity score (results not shown). The impact of ADDHM-Days 1–28 was such that each 10mlkg−1day−1 increase in HM received over the first 28 days post birth translated into a 19% reduction in the odds of sepsis and increases in ADHHM-Days 1–28 were associated with significantly lower NICU costs. Although our findings are consistent with those of previous studies that reported a protective effect of HM feedings for sepsis,6, 7, 8 our study is the first to report an economic impact of ADDHM-Days 1–28 that is attributable to a reduction in the odds of sepsis and hospital costs of care for VLBW infants. Increasing ADDHM-Days 1–28 from <25 to greater than or equal to50mlkg−1day−1 decreased NICU costs by $31514. Even moderate increases in HM dosage from<25 to 25–49.99mlkg−1day−1 were associated with significantly lower costs. This decrease in NICU costs was independently associated with increasing ADDHM-Days 1–28, even after taking infant risk factors and occurrence of sepsis into account.

Our choice of specific doses of 25 and 50mlkg−1day−1 for ADDHM-Days 1–28 was based on previous research by Schanler et al.6 and Furman et al.7 Consistent with our findings, they demonstrated a protective effect of HM for sepsis at threshold doses of greater than or equal to50mlkg−1day−1. However, a major difference between our studies and those of Schanler et al. and Furman et al. is that in both of these studies investigators examined the rates of sepsis as a function of dose, whereas our primary outcome measure was the odds of developing sepsis. Unlike Furman et al., who reported no beneficial effect of ADDHM-Days 1–28 between doses <25 and 25–49mlkg−1day−1, our findings were dose-responsive for these thresholds (Figure 3). Multiple methodological differences between the studies could explain these disparate findings for lower threshold doses.

Several lines of inquiry support the importance of ADDHM-Days 1–28 in reducing the risk of sepsis and other acquired morbidities in premature infants.11 Although it is widely accepted that the bioactive components in HM, including secretory immunoglobulin A, lactoferrin and oligosaccharides,23, 24 provide direct protection from sepsis, the indirect protection afforded by HM during the transition from intrauterine to extrauterine nutrition may be equally important. This is especially true for premature infants, whose gastrointestinal tracts receive less exposure to the gestation-dependent growth factors and protective cytokines provided by swallowed amniotic fluid.25 During the early post-birth period, the immature gut is extremely sensitive to the initial luminal contents.26, 27 Human colostrum has high concentrations of anti-infective and anti-inflammatory components, growth factors and protective cytokines that rapidly grow, mature and protect the immature gut.28 As colostrum transitions to mature milk, high doses of HM facilitate the closure of these paracellular pathways, reducing the risk for translocation of potential pathogens.11, 27 These processes are time-dependent, in that they occur during the critical early post-birth period, which is reflected in ADDHM-Days 1–28.

Of the 175 infants in this study, 57 (33%) received ADDHM-Days 1–28 <25mlkg−1day−1. Our results suggest a substantial cost savings associated with increasing ADDHM-Days 1–28, demonstrating a dose–response relationship between ADDHM-Days 1–28 and onset of sepsis. The incremental cost of sepsis is likely due to several factors. First, infants with sepsis stay longer in the NICU than infants without sepsis. Our results demonstrated that infants with sepsis stayed an average of 28 days longer than those without. In addition to longer NICU stays, treatment of sepsis is resource intensive, due to increased likelihood of ventilation use, which is costly. We estimate that moderately increasing ADDHM to 25–49.99mlkg−1day−1 could have reduced hospital costs by decreasing the odds of sepsis for the infants in this study by $1.2 million ($20384 cost savings multiplied by 57 infants), while increasing the ADDHM for these infants to greater than or equal to50mlkg−1day−1 could have potentially reduced hospital costs by $1.8 million ($31514 cost savings multiplied by 57 infants).

Our results are somewhat higher than the increase in costs due to sepsis in a study by Payne et al.4 of 2809 infants born in 17 NICUs. Payne found that sepsis was associated with a 12% increase in NICU hospitalization costs for infants <1500g. Our results (not shown) demonstrated that sepsis was associated with a 31% increase in NICU hospitalization costs. The study by Payne et al., however, was based on data from 1998–1999. The relative costs of services and supplies used during the NICU hospitalization are likely to have changed over the past decade, hence making comparisons difficult.

The cost savings associated with HM during the NICU hospitalization underestimate the overall impact of HM on societal costs due to the potential for long-term health and neurodevelopmental problems,3 which require ongoing health care and special education services. Future work should extend this line of work to quantify the multiplier effect of cost savings throughout childhood due to a reduction in sepsis during the initial NICU hospitalization. This research team is currently examining this relationship in an economic modeling study.

A potential limitation of our study included the selection of DOL 28 for HM dose assessment, which was at the mean onset of sepsis in our subjects. Although it would be ideal to calculate the HM dose received by each infant before the onset of sepsis, the resulting variation in study interval would make comparisons among sepsis and non-sepsis infants difficult to interpret. Additionally, DOL 28 is a reasonable interval to measure HM dosing as others have also demonstrated a protective effect of HM received during this time interval7 as well as the majority of our infants (84%) had achieved full enteral feedings by DOL 28.

Another potential limitation was that this was a single-center study, which may limit the generalizability of our results. Due to the many lactation initiatives at our NICU, the vast majority of mothers provide at least some HM for their VLBW infants, especially in the early post-birth days (Figure 2). Thus, very few VLBW infants receive only formula, and our HM proportion of enteral intake data are skewed to higher proportions than most NICUs. Therefore, although we did not find a significant difference in HM proportions between sepsis and non-sepsis infants, this may have been due to the high rate of HM feedings at our institution.

The strengths of our study include its prospective design, highly accurate HM dosing data and sample diversity. Our sample reflected the racial and ethnic diversity of VLBW infants cared for in US NICUs,29 and included a representative proportion (47%) of African American infants, a population group that is under-represented in HM research in the United States because the mothers are statistically less likely to provide milk and/or breastfeed.30

In conclusion, our data demonstrate that HM provision of at least 25mlkg−1day−1 over the first 28 DOL was significantly associated with a decrease in sepsis, which is associated with long-term and costly outcomes in VLBW infants.3 The substantial NICU hospital cost savings associated with increased dosages of HM are likely to offset the maternal and institutional costs of HM provision and feeding, such as breast pump rental, lactation care providers and milk storage. This speculation is currently under empirical investigation by this research team.


Conflict of interest

The authors declare no conflict of interest.



  1. Fanaroff AA, Stoll BJ, Wright LL, Carlo WA, Ehrenkranz RA, Stark AR et al. Trends in neonatal morbidity and mortality for very low birthweight infants. Am J Obstet Gynecol 2007; 196(147): e1–147 e8. | Article | PubMed |
  2. Johnson TJ, Patel AL, Jegier B, Engstrom JL, Meier J. The cost of morbidities in very low birth weight infant. J Pediatr 2013; 162: 243–249.e1. | Article | PubMed |
  3. Stoll BJ, Hansen NI, Adams-Chapman I, Fanaroff AA, Hintz SR, Vohr B et al. Neurodevelopmental and growth impairment among extremely low-birth-weight infants with neonatal infection. JAMA 2004; 292: 2357–2365. | Article | PubMed | ISI | CAS |
  4. Payne NR, Carpenter JH, Badger GJ, Horbar JD, Rogowski J. Marginal increase in cost and excess length of stay associated with nosocomial bloodstream infections in surviving very low birth weight infants. Pediatrics 2004; 114: 348–355. | Article | PubMed | ISI |
  5. Stevenson RC, Pharoah POD, Stevenson CJ, McCabe CJ, Cooke RWI. Cost of care for a geographically determined population of low birthweight infants to age 8–9 years. II. children with disability. Arch Dis Child 1996; 74: F118–F121. | Article |
  6. Schanler RJ, Shulman RJ, Lau C. Feeding strategies for premature infants: beneficial outcomes of feeding fortified human milk versus preterm formula. Pediatrics 1999; 103: 1150–1157. | Article | PubMed | ISI | CAS |
  7. Furman L, Taylor G, Minich N, Hack M. The effect of maternal milk on neonatal morbidity of very low-birth-weight infants. Arch Pediatr Adolesc Med 2003; 157: 66–71. | Article | PubMed |
  8. Hylander MA, Strobino DM, Dhanireddy R. Human milk feedings and infection among very low birth weight infants. Pediatrics 1998; 102: E38. | Article | PubMed | CAS |
  9. Patel AL, Meier PP, Engstrom JL. The evidence for use of human milk in very low-birthweight preterm infants. NeoReviews 2007; 8: e459–e466. | Article |
  10. Sisk PM, Lovelady CA, Dillard RG, Gruber KJ, O’Shea TM. Early human milk feeding is associated with a lower risk of necrotizing enterocolitis in very low birth weight infants. J Perinatol 2007; 27: 428–433. | Article | PubMed | ISI | CAS |
  11. Meier PP, Engstrom JL, Patel AL, Jegier BJ, Bruns N. Improving the use of human milk during and after the NICU stay. Clin Perinatol 2010; 37: 217–245. | Article | PubMed | ISI |
  12. Meier PP. Health benefits and cost of human milk for very low birthweight infants 2007; 1, R01-NR010009-01.
  13. Phibbs CS, Schmitt SK. Estimates of the cost and length of stay changes that can be attributed to one-week increases in gestational age for premature infants. Early Hum Dev 2006; 82: 85–95. | Article | PubMed | ISI |
  14. St John EB, Nelson KG, Cliver SP, Bishnoi RR, Goldenberg RL. Cost of neonatal care according to gestational age at birth and survival status. Obstet Gynecol 2000; 182(1 Pt 1): 170–175. | Article |
  15. Rodriguez NA, Meier PP, Groer MW, Zeller JM, Engstrom JL, Fogg L. A pilot study of the oropharyngeal administration of own mother’s colostrum to extremely low birth weight infants. Adv Neonatal Care 2010; 10: 206–212. | Article | PubMed |
  16. Sullivan S, Schanler RJ, Kim JH, Patel AL, Trawoger R, Kiechl-Kohlendorfer U et al. An exclusively human milk-based diet is associated with a lower rate of necrotizing enterocolitis than a diet of human milk and bovine milk-based products. J Pediatr 2010; 156: 562–567. | Article | PubMed | ISI |
  17. Meier PP, Engstrom JL, Murtaugh MA, Vasan U, Meier WA, Schanler RJ. Mothers’ milk feedings in the neonatal intensive care unit: accuracy of the creamatocrit technique. J Perinatol 2002; 22(8): 646–649. | Article | PubMed |
  18. Bureau of Labor Statistics. Consumer price index for all items, 2008–2010. Updated 2012. http://www.bls.gov/data/. Retrieval date 19 February 2012.
  19. Rosenbaum PR, Rubin DB. The central role of the propensity score in observational studies for causal effects. Biometrika 1983; 70: 41–55. | Article | ISI |
  20. Stoll BJ, Hansen N, Fanaroff AA, Wright LL, Carlo WA, Ehrenkranz RA et al. Late-onset sepsis in very low birth weight neonates: the experience of the NICHD neonatal research network. Pediatrics 2002; 110: 285–291. | Article | PubMed | ISI |
  21. Perlman SE, Saiman L, Larson EL. Risk factors for late-onset health care-associated bloodstream infections in patients in neonatal intensive care units. Am J Infect Control 2007; 35: 177–182. | Article | PubMed | ISI |
  22. Manning WG, Mullahy J. Estimating log models: to transform or not to transform? J Health Econ 2001; 20: 461–494. | Article | PubMed | ISI | CAS |
  23. Newburg DS, Walker WA. Protection of the neonate by the innate immune system of developing gut and of human milk. Pediatr Res 2007; 61: 2–8. | Article | PubMed | ISI | CAS |
  24. Lonnerdal B. Nutritional and physiologic significance of human milk proteins. Am J Clin Nutr 2003; 77: 1537S–1543S. | PubMed |
  25. Hirai C, Ichiba H, Saito M, Shintaku H, Yamano T, Kusuda S. Trophic effect of multiple growth factors in amniotic fluid or human milk on cultured human fetal small intestinal cells. J Pediatr Gastroenterol Nutr 2002; 34: 524–528. | Article | PubMed |
  26. Bjornvad CR, Thymann T, Deutz NE, Burrin DG, Jensen SK, Jensen BB et al. Enteral feeding induces diet-dependent mucosal dysfunction, bacterial proliferation, and necrotizing enterocolitis in preterm pigs on parenteral nutrition. Am J Physiol Gastrointest Liver Physiol 2008; 295: G1092–G1103. | Article | PubMed |
  27. Taylor SN, Basile LA, Ebeling M, Wagner CL. Intestinal permeability in preterm infants by feeding type: mother's milk versus formula. Breastfeed Med 2009; 4: 11–15. | Article | PubMed |
  28. Sangild PT. Gut responses to enteral nutrition in preterm infants and animals. Exp Biol Med 2006; 231: 1695–1711.
  29. Stoll B, Hansen N, Bell EF, Shankaran S, Laptook A, Walsh M et al. Neonatal outcomes of extremely preterm infants from the NICHD neonatal research network. Pediatrics 2010; 126: 443–456. | Article | PubMed | ISI |
  30. Li R, Darling N, Maurice E, Barker L, Grummer-Strawn LM. Breastfeeding rates in the United States by characteristics of the child, mother, or family: the 2002 national immunization survey. Pediatrics 2005; 115: e31–e37. | Article | PubMed |


We acknowledge statistical assistance from Michael E Schoeny, PhD and David B Henry, PhD. This study was funded by NIH Grant NR010009.