*2.4. Demographics*

Demographic information was collected for all infants including gender, gestational age at birth and discharge, and day of life (DOL) at discharge. Additional clinical outcomes were collected including the presence of bronchopulmonary dysplasia (BPD), retinopathy of prematurity (ROP), intraventricular hemorrhage (IVH), necrotizing enterocolitis (NEC), diaper dermatitis, and death. Treatment requirements were coded similarly if an infant required: oxygen at 36 weeks estimated gestational age (EGA), ROP procedure, IVH shunt, Avastin treatment, Dexamethasone use, and Bicitra use. ROP stage, IVH grade, and number of days of Dexamethasone use were included if available.

### *2.5. Anthropometrics*

Infants were weighed daily on a gram scale, and head circumference and length (centimeters) were recorded weekly by nursing staff using a measuring tape. Fenton growth curve percentile rankings [18,19] were electronically plotted for each recorded anthropometric measurement. Weight, head circumference, and length measurements with associated Fenton percentile rankings were taken for infants at birth and at 36 weeks EGA, if available.

#### *2.6. Nutrition*

Enteral feeding data collected included day of life (DOL) enteral feedings were initiated, DOL full enteral feedings were reached (with a discontinuation of parenteral nutrition support), and the number of times enteral feedings were held (not secondary to preparation for a procedure). Maximum caloric density and number of days on enteral feedings >24 cal/ounce were collected for infants requiring caloric densities higher than the standard 24 cal/ounce to promote adequate growth.

Daily average provision of calories and protein (g) per kg body weight were calculated for infants in each group if they received 50% of enteral feedings as fortified MBM during NICU stay. These averages were taken when fortified enteral feedings reached a minimum of 140 mL/kg/day until either daily intake was consistently less than this amount, the infant was changed to unfortified MBM, or the infant received greater than 50% infant formula. Growth and nutrition was evaluated for the groups comparing only growth during the period where the infant received 50% of enteral feedings as fortified MBM. An electronic medical system (Intuacare®: Omaha, NE, USA) contained protein references for breast milk and specified enteral formulas and caloric density. Nursing staff documented daily intake of breast milk or specified enteral formulas, thus, daily calorie and protein provision per kg of body weight were electronically calculated using the daily-recorded weight. The electronic medical system also calculated the percentages of MBM *vs*. infant formula received according to nursing documentation.

#### *2.7. Laboratory Measurements*

Maximum creatinine, maximum blood urea nitrogen (BUN) level, maximum base deficit value, maximum calcium level, and lowest carbon dioxide (CO2) lab values were collected, if available, after DOL 14 and DOL 30 for all infants. Values were not collected before DOL 14 to eliminate those reflective of parenteral nutrition support and unfortified enteral feedings. Phosphorus and pH were not consistently nor routinely obtained in this patient population and were therefore not collected in this retrospective study.

#### *2.8. Data Analysis*

Descriptive statistics were displayed for all variables by type of milk (powder *vs*. liquid) given. The Wilcoxon rank sum test was used to compare continuous data between the milks groups. Associations of categorical variables were assessed with the Fisher's exact test. A *p*-value 0.05 was considered statistically significant. To assess the difference in growth patterns between infants given powder and infants given liquid, a mixed effects model was used. We included random slopes and intercepts for each subject to capture individual growth pattern as well as fixed effects for group and day and a group day interaction term. A significant interaction of day and group indicates differing growth patterns based on group. Growth Velocity (GV) was calculated using Equation 1 [20].

$$\text{GV} = \{1000 \times \ln(W\_{\text{n}}/W\_{1})\} (D\_{\text{n}} - D\_{1}) \tag{1}$$

#### **3. Results**

There were 46 infants in the PHMF group (21 males, 25 females) and 23 infants in the ALHMF group (13 males, 10 females) (*p* = 0.45). Additional baseline characteristics were not statistically significant between the two groups, as shown in Table 2. Enteral feeding data, growth and analyzed lab values are displayed in Table 3. Clinical outcomes are displayed in Table 4. ROP stage, IVH grade, and number of days of Dexamethasone use were not statistically significant and are not included in Table 4.


**Table 2**. Baseline characteristics of the subjects.

 Growth at these time points represents nutrition delivery throughout hospitalization not just breast milk with PHMF and ALHMF.


**Table 3**. Enteral feeding, growth and laboratory data.



**Table 4**. Clinical outcomes.

#### *3.1. Safety and Clinical Outcomes*

Mean lowest CO2 lab values (collected while infants were enterally feeding and not acutely ill) were significantly lower in the ALHMF group compared to the PHMF group after both DOL 14 (18.5 *vs*. 23 mmol/L, *p* = 0.002) and DOL 30 (20 *vs*. 25 mmol/L, *p* = 0.002). Lowest CO2 lab values after DOL 14 are displayed comparatively in Figure 1. Lowest values after DOL 30 are displayed similarly in Figure 2. Maximum BUN and creatinine levels were similar between the two fortifier groups and were not statistically significant. All other analyzed lab values were not statistically different. All laboratory data in this retrospective study was obtained for clinical purposes regardless of the fortifier group.

Incidence of NEC (a variable we were not powered to evaluate) was significantly higher in the ALHMF group compared to the PHMF group (13% *vs*. 0%, *p* = 0.03).

**Figure 1.** CO2 levels between groups after Day of Life 14. The lowest CO2 levels after DOL 14 were collected from metabolic panels. The mean level in the powder group was 23, the mean level in the liquid group was 18.5. Laboratory clinical reference range 22–32 mmol/L. The difference is statistically significant (*p* = 0.002).

**Figure 2.** CO2 levels between groups after Day of Life 30. The lowest CO2 levels after DOL 30 were collected from metabolic panels. The mean level in the powder group was 25, the mean level in the liquid group was 20. Laboratory Clinical reference range 22–32 mmol/L. The difference is statistically significant (*p* = 0.002).

#### *3.2. Enteral Nutrition and Growth*

Growth was significantly different between the two groups as measured in g/kg/day and is described in Figure 3. Infant growth as measured in g/day from birth to 36 weeks EGA was 23.7 in the PHMF group and 18.8 in the LHMF group (*p* = 0.057). There were no statistically significant differences in the length of time to full feedings or the number of times that feedings were held that could account for the difference in growth rates between the two groups.

**Figure 3.** The growth pattern of infants receiving powder differs from the growth pattern of infants receiving liquid on fortified feed days. The plot shows the growth pattern for each infant and the fitted line by group. Based on the plot, infants on powder grow at a faster rate than infants receiving liquid. Evaluation of growth in gm/kg/day for the days infants were fed fortified breast milk, based on the mixed effects model, shows a significant interaction between day and group (*p* = 0.0022). Truncating the analysis at 45 days did not attenuate the results.

Daily average protein/kg/day provision was higher in the ALHMF group compared to the PHMF group (4.3 *vs*. 3.9 g, *p* = 0.0014). Mean enteral calorie provisions in the ALHMF group were higher than in the PHMF group, 117.3 kcal/kg/day in the PHMF group as compared to 128.7 kcal/kg/day for infants in the ALHMF group (*p* = 0.057). A higher proportion of infants in the ALHMF group required increased caloric density of feedings >24 cal/ounce as compared to infants in the PHMF group, (48% *vs*. 26%, *p* = 0.10). While this did not reach a statistical difference, clinically this was notable.
