Results of search strategy
Initially, 7163 records were identified. After duplicate removal, 2533 titles and abstracts were screened, of which 61 records were included in full-text screening. Of these, 44 did not meet the inclusion criteria and were excluded; 17 records met the inclusion criteria (Fig. 1), of which 11 studies were included in the review (Table 1) and six studies were identified as ongoing. The characteristics of the six ongoing studies, all RCTs, are described in Supplementary table 2.
Three studies were marked as awaiting classification as we were unable to determine the characteristics of the included population and attempts to contact authors for clarification were unsuccessful.27,28,29
Characteristics of included studies
The characteristics of the included studies are described in Table 1. Included studies were published between 1999 and 2023 and involved 10,147 infants, with sample sizes ranging from 32 to 6209. Two RCTs were identified, one from Finland,30 and the second a pilot study from the USA.31 The remaining studies were eight cohort studies from the USA.15,32,33,34,35,36,37,38 and one cross-sectional study conducted in Finland.39 Only abstracts were available for two studies.32,35
Most of the included studies were conducted in newborn nurseries or postnatal wards.15,30,33,35,36,37,38 In nine studies, supplementation of MOM with DHM was directly compared with IF supplementation.15,30,31,32,33,34,37,38,39 Two studies compared groups before and after protocol changes implementing in-hospital supplementation with DHM.35,36 The gestational age (GA) of participants ranged from 30 weeks to full-term.
Effects of the intervention
Twelve outcomes of interest were reported. Heterogeneity in study design and outcome definitions prevented us from performing meta-analysis. Thus, a summary of studies indicating a potential benefit, an unclear effect, or a possible harm of the use of DHM compared to IF for supplementation of MOM is presented in Table 2.
Primary outcome
Exclusive breastfeeding at hospital discharge
Five studies reported the effects of DHM supplementation on rates of EBF at hospital discharge.15,32,33,35,36 One retrospective cohort study reported that 100% of infants supplemented with human milk in hospital (either expressed breastmilk or DHM) were EBF at hospital discharge, whereas no infants supplemented with IF in hospital were EBF at hospital discharge (p < 0.001).15
Two retrospective cohort studies reported a statistically significant increase in overall EBF rates at discharge following the implementation of DHM in their hospitals.33,35 Alissa et al. 35 reported an increase in yearly average EBF rate from 25% pre-DHM implementation (95% confidence interval [CI] 23.5%–26.5%) to 45.9% post-DHM implementation (95%CI 44.3%- 47.6%, p < 0.0001). Merjaneh et al. 33 reported an increase in average monthly EBF rates at discharge from 33% pre- to 46% post-implementation of DHM (p < 0.005) but did not report confidence intervals or standard deviation. Both studies report data as an overall average percentage of infants who were EBF at discharge, before and after the implementation of DHM and do not report the outcome in the post-DHM implementation cohort separately for infants receiving DHM versus IF supplementation. Thus, the post-DHM implementation group could include both infants who were supplemented with DHM and IF.
The cohort study by Heizelman et al. 32 found no difference in feeding type at hospital discharge between DHM and IF supplementation groups. The cohort study by Ponnapakkam et al. 36 reported an increase in monthly EBF rates at discharge from 33% (95% CI 0–45%) to 55% (95% CI 30–80%) following the introduction of DHM (magnitude of effect estimated from figures); however, rates of breastfeeding at discharge in the post-DHM implementation cohort were not reported by type of supplemental feeding. As there is an overlap in the confidence intervals of pre- and post-DHM supplementation in figures, this difference is likely not statistically significant.
Overall, we judged the direction of findings as of no clear difference between DHM and IF supplementation on EBF rates at hospital discharge (Table 2). Using the GRADE approach, the certainty of evidence for this outcome was judged as very low due to risk of bias, imprecision, inconsistency, and indirectness (Table 3).
Exclusive breastfeeding after hospital discharge
Three studies reported effects of supplementation of MOM with DHM compared to IF on EBF after hospital discharge.33,39 The pilot RCT by Pithia et al. 31 reported similar rates of EBF at 6–8 weeks chronological age in both groups (31% in DHM group and 38% in IF group, p = 0.7). The cross-sectional study by Ikonen et al. 39 reported higher rates of EBF in infants supplemented with DHM compared to those supplemented with IF at 0–1 month (61.1% versus 52.3%, p = 0.2), 2–3 months (60.6% versus 45.1%, p = 0.2) and 4–5 months (40.2% versus 21.9%, p = 0.001).39
Merjaneh et al. 33 reported rates of EBF at < 1 month, 4 months, 4−6 months and ≥ 6 months; however, no group or pair-wise comparison were reported. While EBF at < 1 month was lower in the DHM group compared to the IF group (12% versus 67%, respectively), no direct comparison or p-value was given. The authors reported that EBF at ≥ 6 months was higher in the DHM than the IF group (58% vs 15%, respectively) and infants who had received DHM were five times more likely to be EBF at 6 months of life than those who had received IF (adjusted odds ratio [OR] at 6 months = 5.13, 95% CI 1.37– 19.23, p = 0.01).33 However, there was significant attrition in the IF group (33% versus 46% attrition for DHM and IF groups, respectively).
The available evidence suggests that there was a potential benefit of supplementation of MOM with DHM compared to IF for EBF at 4-6 months, but also no difference for EBF rates up to 3 months of age. Thus, we judged the overall direction of findings as of no clear difference between DHM and IF supplementation on EBF rates after hospital discharge (Table 2). Using the GRADE approach, the certainty of evidence for this outcome was judged as very low due to risk of bias, imprecision, and indirectness (Table 3).
Secondary outcomes
Any breastfeeding at hospital discharge
One retrospective study reported data on any breastfeeding at hospital discharge,15 finding that infants supplemented with IF were significantly less likely to be receiving any breastfeeding at hospital discharge compared to infants supplemented with DHM (Risk ratio 0.84, 95% CI 0.77–0.92). We therefore judged the direction of findings as of possible benefit of DHM on the rates of any breastfeeding at hospital discharge (Table 2).
Any breastfeeding after hospital discharge
Four studies reported data on any breastfeeding after hospital discharge.37,38,39 Gray et al. 37 showed significantly lower rates of breastfeeding at 2 and 6 months postpartum in infants who received IF supplementation in hospital compared to infants who received DHM supplementation (2 months: OR 0.26, 95% CI 0.12−0.56, p = 0.01; 6 months: OR 0.42, 95% CI 0.19−0.94, p = 0.034). Ikonen et al. 39 reported different results according to the time point of outcome measurement. While any breastfeeding rates tended to be higher in the DHM compared to IF supplementation group throughout the first year, this was only statistically significant at 4–5 months (87.4% versus 68.4%, p < 0.001). Riley et al. 38 found that rates of breastfeeding were similar between DHM and IF groups at 1 month postpartum. The pilot RCT by Pithia et al. 31 reported similar rates of infants received a mixed diet (IF and breastmilk) at 6-8 weeks chronological age, with 44% in the DHM supplementation group and 50% in the IF group (p = 0.7).
The effect of DHM supplementation compared to IF on the rates of any breastfeeding after discharge was only significant at some time points and was inconsistent across the three studies reporting this outcome. Thus, we judged the overall direction of findings as of no clear difference between DHM and IF supplementation on any breastfeeding rates after hospital discharge (Table 2).
Hypoglycaemia
Three studies reported outcomes related to hypoglycemia using various definitions.32,34,36 Sen et al. report median change in infants’ blood glucose concentration following buccal dextrose gel plus a DHM or IF feed and found greater blood glucose concentration in infants supplemented with DHM compared to those supplemented with IF, but this was not statistically significant.34 Heizelman investigated rates of hypoglycemia resolution with a single dose of dextrose gel and the number of additional doses required to resolve hypoglycemia and found no difference between study groups.32
Ponnapakkam et al. reported a significant increase in blood glucose concentration following a feed of expressed breastmilk plus supplementation with DHM compared to supplementation with IF (mean (95% CI) increase in blood glucose with DHM: 20.7 (12.7–28.7) mg/dL vs. with IF: 6.6 (2.5–10.7) mg/dL). As there was no overlap in reported confidence intervals, we concluded that the difference reported was likely statistically significant. In addition, this study reports that episodes of symptomatic hypoglycemia remained unchanged following the introduction of DHM in their unit and that time to final hypoglycemic episodes was similar between DHM and IF supplementation.36
Thus, the direction of findings was judged as of no clear difference between DHM and IF supplementation for outcomes relating to hypoglycemia (Table 2).
Infant growth
One study reported outcomes related to infant growth, with no significant difference in mean (SD) weight, length, and head circumference (HC) z-scores between DHM and IF supplementation groups at the end of study intervention.31 At 6–8 weeks chronological age, there was no difference in weight and length of both groups except for higher HC z-score in the IF group (DHM −0.4 (1.3) vs. IF 0.4 (0.9), p = 0.04). Since it was a pilot RCT with small sample size, we judged the overall direction of findings for this outcome as of no clear difference (Table 2). Using the GRADE approach, the certainty of evidence for this outcome was judged as low due to risk of bias and imprecision (Table 3).
Admission to NICU and duration of hospital stay
Ponnapakkam et al. 36 described a reduction in NICU admission rates for treatment of asymptomatic hypoglycaemia from 16% (95% CI 0-38%) in the pre-DHM implementation cohort to 6% (95% CI 0-28%) following the implementation of DHM (confidence intervals estimated from figure). This difference is unlikely to be significant.36 Heizelman et al. found no significant difference in NICU admission rates between DHM and IF supplementation groups; however, it was not reported whether this outcome was a measure of all-cause admission or specifically related to hypoglycemia.32 Mannel et al. 15 reported that length of hospital stay did not differ between study groups.15
Therefore, the overall direction of findings was judged as of no clear difference between DHM and IF supplementation on the rates of admission to NICU and duration of hospital stay (Table 2). Using the GRADE approach, the certainty of evidence for this outcome was judged as very low due to imprecision and indirectness (Table 3).
Childhood morbidity
The study by Saarinen et al. 30 was the only study reporting on childhood morbidity, defined by the authors as the risk of cow’s milk allergy (CMA) during childhood. This RCT found that there was no difference in risk of developing CMA between infants supplemented during hospital stay with cow’s milk-based IF and with DHM. Therefore, the direction of findings was judged to be of no clear difference between DHM and IF supplementation on childhood morbidity (Table 2).
Quality assessment and risk of bias
Among the nine observational studies, we judged five studies to be of weak quality,32,33,35,37,39 two of moderate quality36,38 and two of strong quality15,34 (Fig. 2).
Of the five studies assessed as being of weak quality, four were cohort studies,32,33,35,37 and one was a cross-sectional study.39 The quality of these studies was further downgraded due to: potential for selection bias37,38,40,41; failure to control for confounders37,40; study design37,38,40,41,42; lack or poor description of blinding37,38,40,41,42; data collection methodology,37,38,40 and attrition.38,40
The two studies assessed as of moderate quality were prospective cohort studies.36,38 The quality of these studies was further downgraded due to: lack of control for potential confounders43,44; selection bias43,44; lack or poor description of blinding43,44; data collection methodology,43 and attrition.44
Two studies were assessed as being of strong quality15,34 ; but since both are retrospective cohort studies, they were judged as of moderate quality in the domains of study design, selection bias and blinding. The intervention integrity was judged uncertain within the study by Sen et al. as the authors do not clearly state whether infants were receiving IF or DHM exclusively or as supplementation to breastfeeding.
The quality of the two included RCTs30 was assessed using the ROB-1 tool (Fig. 3). The study by Saarinen et al. 35 was judged to be at low risk of reporting bias due to adequate blinding and concealment of allocation of the intervention from participants, clinicians and outcomes assessors. This study was judged as of unclear risk of bias for random sequence generation as the method was not clearly described, and as high risk of bias for incomplete outcome reporting and other biases. Outcome data were considered as inadequately addressed because, although losses to follow-up were low (0.9%), only 76% of the participants returned completed records of infant feeding regimens, and detection of the primary outcome (diagnosis of CMA) was reliant on parental reporting of infant’s symptoms. Authors assumed that if symptoms were not reported, they were not present; however, it is possible that symptoms were not reported due to attrition. Other biases include selection bias, as baseline characteristics were not described and only 41% of the eligible population agreed to participate in the study, and a potential conflict of interest as this study received funding from industry (Valio Limited, a dairy product manufacturer in Finland; and Nutricia, a brand of Danone specialising in therapeutic food and IF).
The pilot RCT by Pithia et al. 31 was judged to be of low risk of bias in the domains of selective reporting bias, adequate sequence generation, allocation concealment and incomplete outcome data. This study was judged as high risk of bias regarding blinding, as this RCT was unblinded to participants, clinicians and outcome assessors. No other biases were identified.
link