Implementing Feeding Guidelines for NICU Patients <2000 g Results in Less Variability in Nutrition Outcomes 
From * Intermountain Healthcare Neonatology
Research, McKay-Dee Hospital Center, Ogden, Utah; Correspondence: Robert D. Christensen, MD, Intermountain Health Care, 4401 Harrison Blvd, Ogden, UT 84403. Electronic mail may be sent to rdchris4{at}ihc.com. Background: We devised a consistent approach to instituting and advancing enteral nutrition among neonatal intensive care unit (NICU) patients <2000 g birth weight. We then assessed variability in feeding-related outcomes during a period before (period 1) vs after (period 2) implementing these guidelines. Methods: Using data from period 1 vs period 2, we statistically compared the equivalence of variance, focusing on certain feeding-related outcomes. Specific outcomes we chose to examine were (1) day of life when the first enteral feedings were given, (2) number of days during the entire hospitalization when no feedings were given, (3) number of days parenteral nutrition (PN) was administered, and (4) day of life when feedings of 80 mL/k/d and 100 kcal/k/d enteral were achieved. Results: Fifty-eight patients <2000 g were admitted to the NICU in period 1, of which 56 survived to discharge home. In period 2, 68 patients <2000 g were admitted and 66 survived to discharge. Demographic features of the patients in periods 1 and 2 did not differ. In both periods, feedings were begun on a median of day 1. However, in period 1 the range was from day 0 to day 24, and in period 2, the range was from day 0 to day 6 (equivalence of variance p < .001). After feedings were initiated, they were withheld for a median of 2 days (range, 0–23) during the remainder of the hospitalization in period 1 vs a median of 1 day (range, 0–12) in period 2 (p < .001). During period 1, PN was used for a median of 10 days (range, 0–72) vs 7 (range, 0–47) in period 2 (p = .001). During period 1, more variability occurred in the day of life when 80 mL/k/d and 100 kcal/k/d were achieved (both p < .001). No differences were seen in necrotizing enterocolitis, intestinal perforation, mortality, or length of hospital stay. Conclusions: Implementing feeding guidelines was associated with significantly less variability in feeding-related outcomes. We speculate that this is a reflection of better feeding tolerance, which resulted from a more consistent approach to initiating and advancing enteral feedings. Developing consistent approaches to various practices and procedures in neonatology has been suggested as a means of improving outcomes.1–3 As an example, the Vermont Oxford Network "Got Milk" focus group developed guidelines for enteral nutrition that were tested by Kuzma-O'Reilly et al3 and found to improve nutrient intake and growth, with a reduced length of stay and reduced costs. As part of an overall effort to improve outcomes, we used a multidisciplinary consensus development process to devise a consistent approach to enteral nutrition of low-birthweight infants. The multidisciplinary group produced a set of feeding guidelines for neonates <2000 g birth weight according to guidelines originally written (by D.R.G.) at Utah Valley Regional Medical Center in Provo, Utah. The guidelines consisted of instructions regarding the volume of feedings to be administered, the timing of the feedings, the subsequent increases in feeding volumes, and advances in caloric density from 20 to 22 to 24 kcal/oz. The present study was undertaken to assess whether implementing these guidelines was associated with less variability in certain feeding-related outcomes.
Details of the feeding guidelines are posted on the Intermountain Healthcare website. Briefly, the guidelines give specific instructions to the bedside nurses, according to the birth weight of the patient, categorized as <1000 g, 1001–1250 g, 1251–1500 g, or 1501–2000 g. The guidelines were developed in a series of group discussions in September and October 2004. They were tested on a limited number of patients during November and December 2004 and were adopted for general implementation on January 1, 2005. An example of the feeding guideline sheet used for a neonate with a birth weight between 1001 and 1250 g is shown as Table I. Since January 1, 2005, patients admitted to the McKay-Dee Hospital neonatal intensive care unit (NICU) with a birth weight <2000 g have had the birth-weight-appropriate feeding guideline sheet placed at the bedside as part of the bedside nurse charting.
We analyzed data from all patients <2000 g birth weight admitted to the NICU during the 6-month period after implementing these guidelines (date of birth, January 1–June 30, 2005) compared with all patients during the same 6-month period in 2004 (date of birth, January 1–June 30, 2004), prior to beginning discussions about devising and implementing feeding guidelines. Patients in each of the 2 periods were identified from the NICU admission records according to date of birth and birth weight. Demographic information was obtained from the admission log. The feeding volumes administered to each patient each day and the milk or infant formula used were retrieved from the NICU dietary records, as were the NPO days, time to reach 80 mL/k/d, 100 kcal/k/d, and days parenteral nutrition (PN) was used. The Intermountain Healthcare (IHC) institutional review board approved the study. The day of birth was termed "day of life 0," and the day beginning 1 minute after their first midnight was termed "day of life 1." The time to reach 80 mL/k/d was selected as a significant outcome because of an IHC systemwide program to reduce line-associated infections in the NICU, calling for consideration of removing central catheters when 80 mL/k/d of enteral intake is reached.4 PN was defined as an amino-acid-containing, multivitamin-containing, IV solution ordered on the IHC "PN program" and prepared by the hospital pharmacy PN team.
Descriptive statistics were calculated using Stata 8.3 (College Station,
TX). Between-group means were tested using independent-sample t-tests
when parametric assumptions were met, with Wilcoxon ranksum tests used for
nonparametric comparisons. Proportions were compared between groups using
In period 1, 301 patients were admitted to the NICU, of which 58 (19.3%) weighed <2000 g at birth. In period 2, 313 patients were admitted, of which 68 (21.7%) weighed <2000 g. Demographic features of these patients are listed in Table II. No significant differences were observed between patients in the 2 periods in birth weight, gestational age, gender, race, multiple birth rate, or requirement for mechanical ventilation. Fewer patients in period 2 were given feedings with human milk, and there was more use of indomethacin for patent ductus arteriosus closure in period 2 (Table II).
Of the 58 patients admitted to the NICU in period 1, 2 died; a 700-g male infant (23 weeks' gestation) who died on day 0, and a 720-g female infant (26 weeks) who died after 6 weeks. The first of these was not considered in the feeding and growth calculations (Table III), because no feedings were given, but the second was included because that patient survived for 37 days and received enteral feedings. Of the 68 patients in period 2, 2 died; a 657-g male patient (23 wks) and a 990-g female patient (24 weeks). Both died on day of life 1. Neither received any enteral feedings, and neither was included in the feeding-outcome data (Table III).
In periods 1 and 2, the median days to first milk feedings was 1. However, the variability (range, 0–24 days in period 1 vs 0–6 days in period 2) was far less in period 2 (p < .001, Table III). Similarly, during period 1, a median of 2 NPO days occurred between feeding initiation and hospital discharge, and during period 2, this was reduced to a median of 1 day. However, the range of NPO days (0–23 days in period 1 and 0–12 days in period 2) was less during period 2 (p < .001). In period 1, there were more days where PN was used (Table III). Also, more variation was seen in period 1 regarding the day of life when 80 mL/k/d and 100 kcal/k/d were achieved (Table III). No differences were observed between the 2 periods in cases of necrotizing enterocolitis (NEC), intestinal perforation, length of hospital stay, or survival.
During the 6-month period after implementing feeding guidelines, we observed fewer NPO days and fewer days where PN was used. Although these improvements were statistically significant, they were relatively minor improvements compared with those observed by Kuzma-O'Reilly et al3 after they implemented feeding guidelines. They found that adopting feeding guidelines greatly reduced the initial NPO days, improved nutrient intake and growth, and reduced the length of hospital stay. Perhaps one reason we failed to see such marked improvements was that our baseline rates were quite different than theirs. For instance, implementing their guidelines was associated with a reduction in days to start enteral feedings from day of life 8.9 before guidelines to day 4.7 after. In contrast, before instituting guidelines our enteral feeding was begun on (median) day of life 1. Similarly, before guidelines were instituted their neonates required 19 days to achieve an enteral intake of 80 kcal/k/d, and this fell to 6.5 days after. In contrast, before guidelines our neonates required only 7 days to reach an enteral intake of 100 kcal/k/d. Thus, our patients already had very early initiation of feedings and quite rapid escalation of feedings, even before the guidelines were adopted. Although the magnitude of our improvements, after adopting written feeding guidelines, was not as great as those of Kuzma-O'Reilly et al,3 our trends were similar. Perhaps if we had a much larger sample size, the reduction in NPO days and days to achieve various feeding milestones would have been more impressive. However, the fact that we observed less variability in all feeding-related outcomes measured suggests that adopting guidelines can produce benefits even among NICUs that have already instituted early enteral nutrition and relatively rapid feeding escalation practices. We did not identify any adverse effects of adopting these feeding guidelines. For instance, we did not see an increase in NEC, intestinal perforation, or mortality rate. This is consistent with the observations of Schulman et al,5 Berseth,6 Berseth et al,7 and Strodtbeck,8 namely, that earlier administration of enteral feedings does not increase the risk of adverse outcomes. In fact, Patole and de Klerk9 recently reported that implementing feeding guidelines reduced the risk of NEC. Not only did we fail to see adverse effects of adopting feeding guidelines but, to the contrary, like the experience of Premji et al,10 it was our impression that the guidelines constituted a better practice. For instance, the guidelines certainly brought more consistency to our feeding practice, and they clearly saved time for the neonatologists and nurse practitioners because the guidelines practically eliminated the daily writing of feeding orders. Regarding consistency, each of the outcomes we tested had less variability in period 2. Specifically, the range of values was narrower for the day of life first feedings were given, the number of NPO days, the number of days PN was used, and the day of life when 80 mL/k/d and 100 kcal/k/d of milk feedings were given. There is always much inherent variability in NICU feeding outcomes, and at least part of this is according to the heterogeneity of illness among NICU patients. Specifically, NICU patients with more severe and prolonged respiratory difficulties are likely to have slower attainment of enteral feeding milestones, whereas those with uncomplicated respiratory courses are likely to have better tolerance of enteral feedings.11 Despite this inherent variability in NICU feeding outcomes, we found that adopting feeding guidelines reduced the variability in all measured feeding-related outcomes. On that basis, we speculate that using the feeding guidelines was beneficial, producing a more consistent feeding practice and resulting in more consistent feeding outcomes. For reasons that are unclear, fewer mothers provided human milk during period 2. In both periods only 3%–4% of patients received human milk exclusively throughout the entire hospitalization. However, 93% (in period 1) and 77% (in period 2) received human milk either exclusively or partially. Obviously, we must engage in further efforts to make human milk available to our NICU patients. As new ideas come forth to potentially improve feeding practice, nutrition, and growth, we suggest that these ideas should be tested among neonates who are being fed according to written feeding guidelines. The consistency derived from using guidelines can reduce confounding variables that arise when clinicians write daily feeding orders. We speculate that this reduction in variability can facilitate interpretability of research findings. According to our experience, we recommend that NICUs that do not have written feeding guidelines consider the merits of devising or adopting guidelines. Perhaps analogous to guidelines for erythrocyte and platelet transfusions,2 adopting feeding guidelines could improve consistency, allow for more interpretable analysis of outcomes, and help define better NICU practices. The authors thank the neonatologists, neonatal nurse practitioners, NICU staff, and dietary staff at McKay-Dee Hospital Center for their valuable assistance in implementing the feeding guidelines. Received for publication January 20, 2006. Accepted for publication June 30, 2006.
Journal of Parenteral and Enteral Nutrition, Vol. 30, No. 6,
515-518 (2006) This article has been cited by other articles:
|
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
|
|
 |
|
Top
METHODS
RESULTS
2 tests with Yate's continuity correction or, when expected
counts were small, Fisher's exact test. Between-group variances were tested
using independent-samples standard deviation F tests. For demographic
features, two-tailed tests were used. Otherwise, one-tailed tests were
conducted. For all tests, 
was set at .05. 
