Participants

Three studies enrolled preterm or low birth weight infants: Anderson 1979 enrolled preterm infants not expected to receive enteral feeds for five days; van Goudoever 1995 infants at birth weight < 2000 grams; and Xie 2014 infants at < 34 weeks' gestation with birth weight 1000 to 1800 grams.

Most of the remaining trials enrolled very low birth weight or very preterm infants (n = 26): Balasubramanian 2013 enrolled infants with birth weight 900 to 1250 grams; Blanco 2008 birth weight < 1000 grams; Bulbul 2012 at < 32 weeks' gestation; Burattini 2013 birth weight 500 to 1249 grams; Can 2012 at 27 to 33 weeks' gestation; Can 2013 at < 32 weeks' gestation; Clark 2007 at 23 to < 30 weeks' gestation; Heimler 2010 at < 34 weeks' gestation; Ibrahim 2004 birth weight 501 to 1250 grams and at 24 to 32 weeks' gestation; Kashyap 2007 birth weight < 1250 grams; Liu 2015 birth weight 1000 to 2000 grams; Morgan 2014 at < 29 weeks’ gestation and birth weight < 1200 grams; Murdock 1995 birth weight < 2000 grams; Pappoe 2009 birth weight 600 to 1200 grams; Pildes 1973 birth weight < 1500 grams; Rivera 1993 preterm infants with respiratory distress at mean gestation 28.5 weeks; Scattolin 2013 birth weight < 1250 grams; Tan 2008 at < 33 weeks' gestation; Tang 2009 birth weight 1000 to 2000 grams; te Braake 2005 birth weight ≤ 1500 grams; Thureen 2003 birth weight ≤ 1300 grams; Uthaya 2016 at < 31 weeks' gestation; Vaidya 1995 birth weight < 1250 grams; van Lingen 1992 at mean gestation higher group 30.7 weeks and lower group 31.0 weeks; Vlaardingerbroek 2013 birth weight < 1500 grams; and Weiler 2006 at 24 to 32 weeks' gestation.

Two studies enrolled term or near term infants: Makay 2007 enrolled infants at ≥ 35 weeks' gestation whose clinical condition precluded oral feeding for three days; and Hata 2002 enrolled surgical term infants receiving fat‐free parenteral nutrition for 10 days.

Black 1981 enrolled infants admitted for respiratory distress but did not report gestation or birth weight.

Interventions

Outcomes

See Characteristics of included studies for details of outcome reporting for each study. Of the 33 included studies, six were short‐term biochemical studies (Anderson 1979; Murdock 1995; Rivera 1993; Thureen 2003; van Goudoever 1995; van Lingen 1992), one was a trial that enrolled term surgical infants (Hata 2002), and another included infants at > 35 weeks (Makay 2007) ‐ all without substantial clinical reporting. Anderson 1979 reported mean weight loss, mean nitrogen balance, and mean BUN values, but we were not able to use the data in this review. Murdock 1995 reported biochemical tolerance during the first 48 hours, but we were not able to use the data in this review. Pildes 1973 reported mortality, biochemical data, days to regain birth weight, weight gain to 21 days, and time to reach discharge weight but did not report denominators, so we were unable to use the data in this review. Hata 2002 reported cholestasis but no other clinical outcomes. Kashyap 2007 has not yet published data. Of the 21 studies reporting clinical effects that could be included in this review, outcome reporting was variable across studies both for outcomes reported and for timing of reporting. Outcomes reported by more than 50% of studies (≥ 10 of 21) included mortality (14 studies), days to regain birth weight (12 studies), late‐onset sepsis (15 studies), necrotising enterocolitis (14 studies), chronic lung disease (10 studies), and severe intraventricular haemorrhage (11 studies).

We have reported several additional analyses in the primary comparison "Higher versus lower amino acid intake in parenteral nutrition" including weight, length, and head circumference; weight, length, and head circumference z‐scores; patent ductus arteriosus; development quotient scores; nitrogen and protein balance; maximal blood urea nitrogen; hyperkalaemia, and discontinued PN due to biochemical intolerance. As these outcomes were not prespecified, we did not include them in subsequent comparisons and in subgroup analyses.

Primary outcomes

Mortality to discharge (Analysis 1.1; Figure 4): Data show no difference in mortality to discharge (typical RR 0.90, 95% CI 0.69 to 1.17; participants = 1407; studies = 14; I² = 0%; quality of evidence: low). We downgraded quality of evidence owing to imprecision and potential for publication or reporting bias.

Figure 4

---

---

Funnel plot of comparison: 1 Higher versus lower amino acid intake in parenteral nutrition, outcome: 1.1 Mortality to hospital discharge.

Neurodevelopmental disability at 2 years age (Analysis 1.2): Data show no difference in neurodevelopmental disability (typical RR 1.04, 95% CI 0.48 to 2.23; participants = 201; studies = 2; I² = 82% [high]; quality of evidence: very low). Neither study reported a significant difference (te Braake 2005; Vlaardingerbroek 2013). We downgraded quality of evidence owing to risk of bias, inconsistency, imprecision, and potential for publication or reporting bias.

Postnatal growth failure at discharge (weight < 10th percentile) (Analysis 1.3): Data show a reduction in postnatal growth failure (weight < 10th percentile) at discharge (typical RR 0.74, 95% CI 0.56 to 0.97; participants = 203; studies = 3; I² = 22%; RD ‐0.15, 95% CI ‐0.27 to ‐0.02; NNTB 7, 95% CI 4 to 50; quality of evidence: very low). We downgraded quality of evidence owing to risk of bias, imprecision, and potential for publication or reporting bias.

Postnatal growth failure at discharge (weight 2 SD below mean) (Analysis 1.4) (not a prespecified outcome): A single study reported no difference in growth failure (weight 2 standard deviations (SDs) below mean) up at discharge (RR 0.96, 95% CI 0.66 to 1.40; participants = 114) (Burattini 2013).

Postnatal growth failure post discharge (Analysis 1.5) (not a prespecified outcome): A single study reported no difference in growth failure at two years (RR 0.66, 95% CI 0.33 to 1.32; participants = 111) (te Braake 2005).

Secondary outcomes

Days to regain birth weight (Analysis 1.6): Data show a significant reduction in days to regain birth weight (MD ‐1.14, 95% CI ‐1.73 to ‐0.56; participants = 950; studies = 13; I² = 77%; heterogeneity: high).

Maximal weight loss in grams (Analysis 1.7): Data show a reduction in maximal weight loss in grams (MD ‐22.71 g, 95% CI ‐33.68 to ‐11.74; participants = 235; studies = 3; I² = 81%; heterogeneity: high).

Weight loss per cent (Analysis 1.8): Data show no difference in weight loss per cent (MD ‐0.33%, 95% CI ‐1.61 to 0.96; participants = 288; studies = 4; I² = 38%; heterogeneity: low). The two meta‐analyses of weight loss outcomes comprise different studies.

Weight gain to 1 month age (Analysis 1.9): Data show a reduction in weight gain to one month of age (MD ‐1.50 g/kg/d, 95% CI ‐2.56 to ‐0.44; participants = 373; studies = 4; I² = 87%; heterogeneity: high).

Weight gain to discharge (Analysis 1.9): Data show no difference in weight gain up to discharge (MD 0.76 g/kg/d, 95% CI ‐0.02 to 1.54; participants = 291; studies = 4; I² = 0%; quality of evidence: very low). We downgraded quality of evidence owing to risk of bias, imprecision, and potential for publication or reporting bias.

Linear growth to 1 month (Analysis 1.10): Data show a reduction in linear growth to one month (MD ‐0.16 cm/week, 95% CI ‐0.26 to ‐0.06; participants = 245; studies = 2; I² = 86%; heterogeneity: high).

Head circumference growth to one month (Analysis 1.11): Data show no difference in head circumference growth to one month (MD 0.01 cm/week, 95% CI ‐0.04 to 0.06; participants = 476; studies = 4; I² = 92%; heterogeneity: high).

Head circumference growth to discharge (Analysis 1.11): Data show an increase in head circumference growth to discharge (MD 0.09 cm/week, 95% CI 0.06 to 0.13; participants = 315; studies = 4; I² = 90%; heterogeneity: high; quality of evidence: very low). We downgraded quality of evidence owing to risk of bias, inconsistency, imprecision, and potential for publication or reporting bias.

Weight change z‐score to 1 month (Analysis 1.12): A single study reported no difference in weight change z‐score to one month (MD ‐0.20, 95% CI ‐0.62 to 0.22; participants = 96) (Vlaardingerbroek 2013).

Weight change z‐score to discharge (Analysis 1.12): Data show no difference in weight change in z‐score to discharge (MD 0.01, 95% CI ‐0.33 to 0.36; participants = 207; studies = 2; I² = 48%; heterogeneity: low).

Weight change z‐score post discharge (Analysis 1.12): Data show no difference in weight change in z‐score post discharge (MD 0.13, 95% CI ‐0.26 to 0.52; participants = 201; studies = 2; I² = 47%; heterogeneity: low).

Length change z‐score: This was not reported.

Head circumference change in z‐score to 1 month (Analysis 1.13): Data show an increase in head circumference change in z‐score to one month (MD 0.27, 95% CI 0.08 to 0.46; participants = 231; studies = 2; I² = 66%).

Head circumference change in z‐score to discharge (Analysis 1.13): Data show no difference in head circumference change in z‐score to discharge (MD 0.18, 95% CI ‐0.15 to 0.50; participants = 207; studies = 2; I² = 63%; heterogeneity: moderate).

Head circumference change in z‐score post discharge (Analysis 1.13): Data show no difference in head circumference change in z‐score post discharge ((MD 0.25, 95% CI ‐0.14 to 0.64; participants = 201; studies = 2; I² = 50%).

Weight at one month (Analysis 1.14) (not a prespecified outcome): Data show no difference in weight at one month (MD ‐18.45 g/kg/d, 95% CI ‐68.42 to 31.52; participants = 430; studies = 4; I² = 78%; heterogeneity: high).

Weight at discharge (Analysis 1.14) (not a prespecified outcome): Data show an increase in weight at discharge (MD 81.07 g/kg/d, 95% CI 36.59 to 125.56; participants = 874; studies = 10; I² = 0%).

Weight post discharge (Analysis 1.14) (not a prespecified outcome): Data show no difference in weight post discharge (MD ‐11.07 g, 95% CI ‐493.31 to 471.18; participants = 211; studies = 2; I² = 0%).

Length at one month (Analysis 1.15) (not a prespecified outcome): Data show no difference in length at one month (MD ‐0.41 cm, 95% CI ‐1.03 to 0.20; participants = 295; studies = 3; I² = 76%; heterogeneity: high).

Length at discharge (Analysis 1.15) (not a prespecified outcome): Data show an increase in length at discharge (MD 0.57 cm, 95% CI 0.17 to 0.98; participants = 553; studies = 6; I² = 47%; heterogeneity: low).

Length post discharge (Analysis 1.15) (not a prespecified outcome): A single study reported no difference in length post discharge (MD ‐0.10 cm, 95% CI ‐1.81 to 1.61; participants = 100) (Burattini 2013).

Head circumference at one month (Analysis 1.16) (not a prespecified outcome): Data show no difference in head circumference to one month (MD 0.19 cm, 95% CI ‐0.13 to 0.51; participants = 430; studies = 4; I² = 81%; heterogeneity: high).

Head circumference at discharge (Analysis 1.16) (not a prespecified outcome): Data show no difference in head circumference at discharge (MD 0.08, 95% CI ‐0.14 to 0.29; participants = 834; studies = 9; I² = 59%; heterogeneity: moderate).

Head circumference post discharge (Analysis 1.16) (not a prespecified outcome): Data show no difference in head circumference post discharge (MD ‐0.04, 95% CI ‐0.52 to 0.44; participants = 211; studies = 2; I² = 3%).

Weight z‐score at one month (Analysis 1.17) (not a prespecified outcome): A single study reported no difference in weight z‐score at one month (MD 0.14, 95% CI ‐0.11 to 0.39; participants = 135) (Morgan 2014).

Weight z‐score at discharge (Analysis 1.17) (not a prespecified outcome): Data show no difference in weight z‐score at discharge (MD 0.16, 95% CI ‐0.02 to 0.33; participants = 352; studies = 3; I² = 0%).

Length z‐score at discharge (Analysis 1.18) (not a prespecified outcome): Data show no difference in length z‐score at discharge (MD 0.12, 95% CI ‐0.14 to 0.38; participants = 228; studies = 2; I² = 0%).

Head circumference z‐score at one month (Analysis 1.19) (not a prespecified outcome): A single study reported an increase in head circumference z‐score to one month (MD 0.30, 95% CI 0.01 to 0.59; participants = 135) (Morgan 2014).

Head circumference z‐score at discharge (Analysis 1.19) (not a prespecified outcome): Data show no difference in head circumference z‐score at discharge (MD 0.04, 95% CI ‐0.18 to 0.26; participants = 354; studies = 3; I² = 57%; heterogeneity: moderate).

Head circumference z‐score at post discharge (Analysis 1.19) (not a prespecified outcome): A single study reported no difference in head circumference z‐score at post discharge (MD ‐0.01, 95% CI ‐0.50 to 0.48; participants = 100) (Burattini 2013).

Days to full enteral feeds (Analysis 1.20): Data show no difference in days to full enteral feeds (MD ‐0.19, 95% CI ‐1.07 to 0.70; participants = 778; studies = 11; I² = 46%; heterogeneity: low).

Late‐onset sepsis (Analysis 1.21): Data show no difference in late‐onset sepsis (typical RR 0.96, 95% CI 0.79 to 1.18; participants = 1255; studies = 15; I² = 0%).

Necrotising enterocolitis (Analysis 1.22): Data show no difference in necrotising enterocolitis (typical RR 1.00, 95% CI 0.68 to 1.47; participants = 1301; studies = 14; I² = 0%).

Chronic lung disease (Analysis 1.23): Data show no difference in chronic lung disease (typical RR 1.04, 95% CI 0.89 to 1.23; participants = 819; studies = 10; I² = 22%).

Patent ductus arteriosus (Analysis 1.24) (not a prespecified outcome): Data show no difference in patent ductus arteriosus (typical RR 0.83, 95% CI 0.67 to 1.02; participants = 607; studies = 7; I² = 13%).

Intraventricular haemorrhage (Analysis 1.25): Data show no difference in intraventricular haemorrhage (typical RR 1.12, 95% CI 0.74 to 1.69; participants = 341; studies = 3; I² = 0%).

Severe intraventricular haemorrhage (Analysis 1.26): Data show no difference in severe intraventricular haemorrhage (typical RR 1.16, 95% CI 0.74 to 1.82; participants = 904; studies = 11; I² = 0%).

Periventricular leukomalacia (Analysis 1.27): Data show no difference in periventricular leukomalacia (typical RR 0.55, 95% CI 0.24 to 1.25; participants = 720; studies = 7; I² = 13%).

MRI brain abnormality at term: No data were reported.

Retinopathy of prematurity (Analysis 1.28): Data show a reduction in retinopathy of prematurity (typical RR 0.44, 95% CI 0.21 to 0.93; participants = 269; studies = 4; I² = 31%; heterogeneity: low).

Severe retinopathy of prematurity (Analysis 1.29): Data show no difference in severe retinopathy of prematurity (typical RR 0.96, 95% CI 0.56 to 1.63; participants = 672; studies = 8; I² = 34%; heterogeneity: low; quality of evidence: very low). We downgraded quality of evidence owing to risk of bias, imprecision, and potential for publication or reporting bias.

Cerebral palsy (Analysis 1.30): Data show no difference in cerebral palsy (typical RR 4.00, 95% CI 0.89 to 17.97; participants = 122; studies = 2; I² = 0%).

Developmental delay (Analysis 1.31): Data show no difference in developmental delay (typical RR 1.35, 95% CI 0.52 to 3.53; participants = 301; studies = 3; I² = 0%).

Blindness (Analysis 1.32): Data show no difference in blindness (typical RR 2.00, 95% CI 0.20 to 19.91; participants = 122; studies = 2; I² = 0%).

Deafness (Analysis 1.33): A single study reported no severe hearing impairment at two years in either group (Vlaardingerbroek 2013).

Bayley Mental Development Index (MDI) at ≥ 18 months (Analysis 1.34): Data show no difference in Bayley MDI at ≥ 18 months (MD ‐4.18, 95% CI ‐8.53 to 0.17; participants = 105; studies = 2; I² = 0%).

Bayley Psychomotor Development Index (PDI) at ≥ 18 months (Analysis 1.36): A single study reported no difference in Bayley PDI ≥ 18 months (MD 3.00, 95% CI ‐6.41 to 12.41; participants = 32) (Blanco 2008).

Bayley III score ≥ 18 months (Analysis 1.35): A single study reported no difference in Bayley III score ≥ 18 months (MD 3.00, 95% CI ‐2.52 to 8.52; participants = 100) (Burattini 2013).

Autism (Analysis 1.37) (not a prespecified outcome): A single study reported no difference in autism (RR 1.00, 95% CI 0.07 to 14.64; participants = 32) (Blanco 2008).

Nitrogen balance (Analysis 1.38): Data show an increase in nitrogen balance (MD 505.20 mg/kg/d, 95% CI 492.01 to 518.39; participants = 153; studies = 6; I² = 99%; heterogeneity: high). Studies are ordered by initial amino acid intake. We noted significant subgroup differences (P < 0.00001).

Protein balance (Analysis 1.39) (not a prespecified outcome): Data show an increase in protein balance (MD 1.57 g/kg/d, 95% CI 1.47 to 1.66; participants = 52; studies = 3; I² = 93%; heterogeneity: high). Studies are ordered by initial amino acid intake. We noted significant subgroup differences (P < 0.00001).

Abnormal serum ammonia (Analysis 1.40): A single study reported no difference in ammonia > 69 μmol/L (RR 13.42, 95% CI 0.79 to 228.24; participants = 61) and ammonia > 122 μmol/L (RR 3.10, 95% CI 0.13 to 73.16; participants = 61) (Blanco 2008). Data show no difference in ammonia > 100 μmol/L (typical RR 9.29, 95% CI 0.52 to 165.45; participants = 105; studies = 2; I² = 0%). All infants with high ammonia were included in the higher amino acid group.

Abnormal blood urea nitrogen (Analysis 1.41): Criteria for abnormal blood urea nitrogen differed between studies. Data show a significant increase in abnormal blood urea nitrogen level (various criteria) (RD 0.26, 95% CI 0.20 to 0.32; participants = 688; studies = 7; I² = 90%; heterogeneity: high; RD 0.26, 95% CI 0.20 to 0.32; NNTH 4; 95% CI 3 to 5).

Maximal blood urea nitrogen (Analysis 1.42) (not a prespecified outcome): Data show an increase in maximal blood urea nitrogen (MD 4.48, 95% CI 3.43 to 5.53; participants = 159; studies = 2; I² = 96%; heterogeneity: high; quality of evidence: high).

Hyperglycaemia (plasma glucose > 8.3 mmol/L) (Analysis 1.43): Data show a reduction in hyperglycaemia (typical RR 0.69, 95% CI 0.49 to 0.96; participants = 505; studies = 5; I² = 68%; heterogeneity: moderate).

Hyperglycaemia treated with insulin (Analysis 1.44): Data show no difference in hyperglycaemia treated with insulin (typical RR 1.24, 95% CI 0.93 to 1.66; participants = 534; studies = 5; I² = 67%; heterogeneity: moderate).

Hypoglycaemia (Analysis 1.45): Data show no difference in hypoglycaemia (typical RR 1.17, 95% CI 0.84 to 1.63; participants = 376; studies = 3; I² = 0%).

Hypoalbuminaemia: This was not reported.

Metabolic acidosis (Analysis 1.46): Data show no difference in metabolic acidosis (typical RR 2.05, 95% CI 0.94 to 4.47; participants = 305; studies = 4; I² = 22%).

Cholestasis (Analysis 1.47): Data show no difference in cholestasis (typical RR 1.26, 95% CI 0.86 to 1.84; participants = 616; studies = 5; I² = 8%).

Hyperkalaemia (Analysis 1.48) (not a prespecified outcome): A single study reported no difference in hyperkalaemia (RR 0.62, 95% CI 0.16 to 2.37; participants = 61) (Blanco 2008).

Discontinued PN due to biochemical intolerance (Analysis 1.49) (not a prespecified outcome): A single study reported no difference in discontinued PN due to biochemical intolerance (RR 13.42, 95% CI 0.79 to 228.24; participants = 61) (Blanco 2008). All six infants who discontinued PN were in the higher amino acid group.

Subgroup analyses

No data were available for review authors to determine the effect of higher versus lower amino acid intake in parenteral nutrition subgrouped by studies enrolling 'sick' infants (e.g. infants with moderate‐severe respiratory distress, receiving cardiovascular support, with possible sepsis, acidosis); and studies enrolling 'surgical' or postoperative infants or infants post cardiopulmonary bypass. The following analyses substantially relate to preterm and low birth weight infants.

Primary outcomes

Mortality to discharge (Analysis 2.1): Data show no difference in mortality (typical RR 0.78, 95% CI 0.45 to 1.36; participants = 433; studies = 6) and no subgroup differences by commencement intake; testing for subgroup differences: P = 0.79, I² = 0%.

Neurodevelopmental disability at 2 years (Analysis 2.2): Data show no difference in neurodevelopmental disability at two years (typical RR 1.04, 95% CI 0.48 to 2.23; participants = 201; studies = 2); testing for subgroup differences was significant: P = 0.02, I² = 81.2%. Neither study reported a significant difference (high AA intake > 2 to ≤ 3 g/kg/d ‐ te Braake 2005; very high AA intake > 3 g/kg/d ‐ Vlaardingerbroek 2013).

Postnatal growth failure at discharge (Analysis 2.3): Data show a reduction in postnatal growth failure at discharge (typical RR 0.74, 95% CI 0.56 to 0.97; participants = 203; studies = 3) and no subgroup differences by commencement intake; testing for subgroup differences: P = 0.21, I² = 37.6%. For subgroups, a single study commencing low amino acid intake (> 1 to ≤ 2 g/kg/d) among infants reported no difference in growth failure at discharge (RR 0.95, 95% CI 0.62 to 1.46; participants = 42) (Pappoe 2009). Infants commenced on high amino acid intake (2 to ≤ 3 g/kg/d) showed a reduction in postnatal growth failure at discharge (typical RR 0.67, 95% CI 0.48 to 0.94; participants = 161; studies = 2).

Postnatal growth failure post discharge (Analysis 2.4): A single study commencing high amino acid intake (> 2 to ≤ 3 g/kg/d) reported no difference in postnatal growth failure at two years (RR 0.66, 95% CI 0.33 to 1.32; participants = 111) (te Braake 2005).

Secondary outcomes

Days to regain birth weight (Analysis 2.5): Data show no difference in days to regain birth weight (MD 0.43, 95% CI ‐0.51 to 1.37; participants = 303; studies = 6) and no subgroup difference by commencement intake; testing for subgroup differences: P = 0.06, I² = 64.2%.

Maximal weight loss in grams (Analysis 2.6): A single study commencing with high amino acid intake (> 2 to ≤ 3 g/kg/d) reported no difference in maximal weight loss in grams (MD 22.60, 95% CI ‐7.25 to 52.45; participants = 50) (Can 2012).

Maximal weight loss per cent (Analysis 2.7): Data show no difference in maximal weight loss per cent (MD ‐2.73, 95% CI ‐5.71 to 0.25; participants = 59; studies = 2; I² = 40%). Both studies commenced with low amino acid intake (> 1 to ≤ 2 g/kg/d).

Weight gain to one month (Analysis 2.8): Data show reduced weight gain to one month (MD ‐3.17 g/kg/d, 95% CI ‐4.49 to ‐1.84; participants = 219; studies = 2) with significant subgroup differences; testing for subgroup differences: P = 0.01, I² = 83.3%. For subgroups, Balasubramanian 2013 commencing 3 g/kg/d versus 1 g/kg/d amino acid on day 1 reported reduced weight gain to one month of age (MD ‐4.48, 95% CI ‐6.17 to ‐2.79; participants = 123). Vlaardingerbroek 2013 commencing with 3.6 g/kg/d versus 2.4 g/kg/d at day 2 to 6 reported no difference in weight gain to one month (MD ‐1.10, 95% CI ‐3.22 to 1.02; participants = 96).

Weight gain to discharge (Analysis 2.9): Data show no difference in weight gain to discharge (MD 1.05, 95% CI ‐0.55 to 2.66; participants = 140; studies = 2) and no subgroup difference by commencement intake; testing for subgroup differences: P = 0.34; I² = 0%.

Linear growth to one month (Analysis 2.10): A single study commencing 3 g/kg/d versus 1 g/kg/d amino acid on day 1 reported a reduction in linear growth to one month (MD ‐0.27, 95% CI ‐0.40 to ‐0.14; participants = 123) (Balasubramanian 2013).

Head circumference growth to one month (Analysis 2.11): Data show reduced head circumference growth to one month (MD ‐0.12, 95% CI ‐0.21 to ‐0.04; participants = 219; studies = 2) with significant subgroup differences: P < 0.0001; I² = 94.9%. For subgroups, Balasubramanian 2013 commencing 3 g/kg/d versus 1 g/kg/d amino acid on day 1 reported reduced head circumference growth to one month of age (MD ‐0.38, 95% CI ‐0.51 to ‐0.24; participants = 123). Vlaardingerbroek 2013 commencing 3.6 g/kg/d versus 2.4 g/kg/d on day 2 to 6 reported no difference in head circumference growth to one month (MD 0.02, 95% CI ‐0.09 to 0.13; participants = 96).

Head circumference growth to discharge (Analysis 2.12): A single study commencing 3.6 g/kg/d versus 2.4 g/kg/d on day 2 to 6 reported no difference in head circumference growth to discharge (MD 0.03, 95% CI ‐0.03 to 0.09; participants = 96) (Vlaardingerbroek 2013).

Weight gain change z‐score up to one month (Analysis 2.13): A single study commencing 3.6 g/kg/d versus 2.4 g/kg/d on day 2 to 6 reported no difference in weight gain change z‐score up to one month (MD ‐0.20, 95% CI ‐0.62 to 0.22; participants = 96) (Vlaardingerbroek 2013).

Weight gain change z‐score to discharge (Analysis 2.14): Data show no difference in weight gain change z‐score up to discharge (MD 0.01, 95% CI ‐0.33 to 0.36; participants = 207; studies = 2) and no subgroup difference by commencement intake; testing for subgroup differences: P = 0.17; I² = 48%.

Weight gain change z‐score post discharge (Analysis 2.15): Data show no difference in weight gain change z‐score post discharge (MD 0.13, 95% CI ‐0.26 to 0.52; participants = 201; studies = 2) and no subgroup difference by commencement intake; testing for subgroup differences: P = 0.17, I² = 47.1%.

Head circumference change z‐score up to one month (Analysis 2.16): A single study commencing 3.6 g/kg/d versus 2.4 g/kg/d on day 2 to 6 reported no difference in head circumference change z‐score up to one month (MD 0.00, 95% CI ‐0.36 to 0.36; participants = 96) (Vlaardingerbroek 2013).

Head circumference change z‐score to discharge (Analysis 2.17): Data show no difference in head circumference change z‐score up to discharge (MD 0.18, 95% CI ‐0.15 to 0.50; participants = 207; studies = 2) and no subgroup difference by commencement intake; testing for subgroup differences: P = 0.10, I² = 62.7%. Neither study reported a significant effect on head circumference change z‐score to discharge.

Head circumference change z‐score post discharge (Analysis 2.18): Data show no difference in head circumference change z‐score up to discharge (MD 0.25, 95% CI ‐0.14 to 0.64; participants = 201; studies = 2) and no subgroup difference by commencement intake; testing for subgroup differences: P = 0.16, I² = 50.1%.

Days to full enteral feeds (Analysis 2.19): Data show no difference by commencement intake in days to full enteral feeds (MD ‐0.22, 95% CI ‐1.60 to 1.17; participants = 196; studies = 4); testing for subgroup differences: P = 0.88, I² = 0%.

Late‐onset sepsis (Analysis 2.20): Data show no difference in late‐onset sepsis (typical RR 0.94, 95% CI 0.65 to 1.38; participants = 319; studies = 5) and no subgroup difference by commencement intake; testing for subgroup differences: P = 0.93, I² = 0%.

Necrotising enterocolitis (Analysis 2.21): Data show no difference in necrotising enterocolitis (RR 0.96, 95% CI 0.45 to 2.03; participants = 340; studies = 5) and no subgroup difference by commencement intake; testing for subgroup differences: P = 0.18, I² = 44%.

Chronic lung disease (Analysis 2.22): Data show no difference in chronic lung disease (RR 1.32, 95% CI 0.86 to 2.02; participants = 202; studies = 4) and no subgroup difference by commencement intake; testing for subgroup differences: P = 0.90, I² = 0%.

Patent ductus arteriosus (Analysis 2.23): Data show no difference in patent ductus arteriosus (RR 0.73, 95% CI 0.50 to 1.07; participants = 244; studies = 4) and no subgroup difference by commencement intake; testing for subgroup differences: P = 0.11, I² = 54.9%. Meta‐analysis of two studies on high amino acid intake (> 2 to ≤ 3 g/kg/d) found a reduction in patent ductus arteriosus (RR 0.42, 95% CI 0.20 to 0.89; participants = 173; studies = 2) (Balasubramanian 2013; Can 2012), whereas the other studies reported no difference from low amino acid intake (2 g/kg/d) ‐ Pappoe 2009 (RR 1.05, 95% CI 0.63 to 1.74; participants = 42) ‐ or very high amino acid intake (3.5 g/kg/d) ‐ Ibrahim 2004 (RR 1.07, 95% CI 0.50 to 2.28; participants = 29).

Intraventricular haemorrhage (Analysis 2.24): A single study reported no difference in intraventricular haemorrhage (RR 1.26, 95% CI 0.41 to 3.91; participants = 123) (Balasubramanian 2013).

Severe intraventricular haemorrhage (Analysis 2.25): Data show no difference in severe intraventricular haemorrhage (RR 1.44, 95% CI 0.66 to 3.17; participants = 261; studies = 5) and no subgroup difference by commencement intake; testing for subgroup differences: P = 0.59, I² = 0%.

Periventricular leukomalacia (Analysis 2.26): Data show only one infant with periventricular leukomalacia in the lower amino acid group (two studies, 146 infants).

Retinopathy of prematurity (Analysis 2.27): Data show a reduction in retinopathy of prematurity (RR 0.44, 95% CI 0.21 to 0.93; participants = 269; studies = 4) and no subgroup difference by commencement intake; testing for subgroup differences: P = 0.76, I² = 0%. Meta‐analysis of studies commencing with high amino acid intake (> 2 to ≤ 3 g/kg/d) found a reduction in retinopathy of prematurity (RR 0.36, 95% CI 0.14 to 0.95; participants = 198; studies = 2). A single trial commencing at low amino acid intake (> 1 to ≤ 2 g/kg/d) reported no difference in retinopathy of prematurity (RR 0.55, 95% CI 0.10 to 2.96; participants = 42) (Pappoe 2009). A single trial commencing at very high amino acid intake (3.5 g/kg/d) also reported no difference (RR 0.71, 95% CI 0.14 to 3.66; participants = 29) (Ibrahim 2004).

Severe retinopathy of prematurity (Analysis 2.28): Data show no difference in severe retinopathy of prematurity (RR 0.47, 95% CI 0.20 to 1.11; participants = 265; studies = 4) and no subgroup difference by commencement intake; testing for subgroup differences: P = 0.17, I² = 43%. Meta‐analysis of high amino acid intake (3 g/kg/d vs 1.5 g/kg/d) at commencement found a reduction in severe retinopathy of prematurity (RR 0.23, 95% CI 0.06 to 0.85; participants = 125; studies = 2). Pappoe 2009 reported no difference from low amino acid intake (2 g/kg/d vs 1 g/kg/d) (RR 0.55, 95% CI 0.10 to 2.96; participants = 42), and Vlaardingerbroek 2013 reported no difference from very high amino acid intake (3.6 g/kg/d vs 2.4 g/kg/d) (RR 5.00, 95% CI 0.25 to 101.53; participants = 98).

Cerebral palsy (Analysis 2.29): A single study commencing 3.6 g/kg/d versus 2.4 g/kg/d on day 2 to 6 reported no difference in cerebral palsy (RR 5.00, 95% CI 0.61 to 41.11; participants = 90) (Vlaardingerbroek 2013).

Developmental delay (Analysis 2.30): Data show no difference in developmental delay at ≥ 18 months (RR 1.04, 95% CI 0.35 to 3.11; participants = 201; studies = 2); and no subgroup difference by commencement intake; testing for subgroup differences: P = 0.53, I² = 0%.

Blindness (Analysis 2.31): No infant was reported as being blind in a single study (Vlaardingerbroek 2013).

Deafness (Analysis 2.32): No infant was reported as being deaf in a single study (Vlaardingerbroek 2013).

Abnormal serum ammonium (Analysis 2.33): A single study commencing on 3 g/kg/d amino acid intake reported that none of 42 infants had a serum ammonia > 100 μmol/L (Bulbul 2012).

Abnormal blood urea nitrogen (Analysis 2.34): Data show an increase in abnormal blood urea nitrogen (RR 2.11, 95% CI 1.44 to 3.08; participants = 138; studies = 2) and no subgroup difference by commencement intake; testing for subgroup differences: P = 0.88, I² = 0%. Individual studies reported no difference in BUN > 14.3 mmol/L among infants commenced on 2 g/kg/d ‐ Pappoe 2009 (RR 2.48, 95% CI 0.28 to 21.93; participants = 42) ‐ and an increase in BUN > 10 mmol/L among infants commencing on 3.6 g/kg/d ‐ Vlaardingerbroek 2013 (RR 2.09, 95% CI 1.43 to 3.04; participants = 96).

Hyperglycaemia (Analysis 2.35): A single study of infants receiving very high amino acid intake (> 3 g/kg/d) reported no difference in hyperglycaemia (RR 1.47, 95% CI 0.85 to 2.53; participants = 42) (Pappoe 2009).

Hyperglycaemia treated with insulin (Analysis 2.36): Data show an increase in hyperglycaemia treated with insulin of borderline significance (RR 1.84, 95% CI 0.97 to 3.49; participants = 138; studies = 2) and a subgroup difference of borderline significance by commencement intake; testing for subgroup differences: P = 0.07, I² = 70%. A single study reported an increase in infants commencing on low amino acid intake at 2 g/kg/d versus 1 g/kg/d (RR 4.96, 95% CI 1.26 to 19.47; participants = 42) (Pappoe 2009). Another single study reported no difference among infants commencing on very high amino acid intake (3.6 g/kg/d vs 2.4 g/kg/d) (RR 1.15, 95% CI 0.54 to 2.45; participants = 96) (Vlaardingerbroek 2013).

Hypoglycaemia (Analysis 2.37): A single study reported no difference in hypoglycaemia among infants commencing on high amino acid intake at 3 g/kg/d (RR 1.68, 95% CI 0.83 to 3.41; participants = 123) (Balasubramanian 2013).

Metabolic acidosis (Analysis 2.38): No infant was reported as having metabolic acidosis in a single study commencing on low amino acid intake at 1.15 g/kg/d (van Goudoever 1995).

Primary outcomes

Mortality (Analysis 3.1): Data show no difference in mortality (RR 0.94, 95% CI 0.57 to 1.55; participants = 292; studies = 2).

Neurodevelopmental disability and postnatal growth failure: Trials provided no data.

Secondary outcomes

Head circumference growth to one month (Analysis 3.2): Morgan 2014 reported an increase in head circumference growth to one month (MD 0.13 cm/week, 95% CI 0.05 to 0.20; participants = 135).

Head circumference change z‐score to one month (Analysis 3.3): Morgan 2014 reported an increase in head circumference change z‐score to one month (MD 0.37, 95% CI 0.15 to 0.59; participants = 135).

Days to regain birth weight (Analysis 3.4): Tan 2008 reported a reduction in days to regain birth weight (MD ‐3.60 days, 95% CI ‐5.88 to ‐1.32; participants = 114).

Days to full enteral feeds (Analysis 3.5): Tan 2008 reported an increase in days to full enteral feeds (MD 4.00 days, 95% CI 1.01 to 6.99; participants = 114).

Late‐onset sepsis (Analysis 3.6): Morgan 2014 reported no difference in late‐onset sepsis (RR 0.94, 95% CI 0.63 to 1.41; participants = 127).

Necrotising enterocolitis (Analysis 3.7): Data show no difference in necrotising enterocolitis (typical RR 0.76, 95% CI 0.37 to 1.59; participants = 241; studies = 2).

Chronic lung disease (Analysis 3.8): Data show no difference in chronic lung disease (typical RR 1.10, 95% CI 0.92 to 1.31; participants = 241; studies = 2).

Patent ductus arteriosus (Analysis 3.9): Morgan 2014 reported no difference in patent ductus arteriosus (RR 1.02, 95% CI 0.66 to 1.56; participants = 127).

Severe intraventricular haemorrhage (Analysis 3.10): Data show no difference in severe intraventricular haemorrhage (typical RR 1.16, 95% CI 0.51 to 2.63; studies = 2).

Periventricular leukomalacia (Analysis 3.11): Morgan 2014 reported no difference in periventricular leukomalacia (RR 2.03, 95% CI 0.39 to 10.70; participants = 127).

Severe retinopathy of prematurity (Analysis 3.12): Morgan 2014 reported no difference in severe retinopathy of prematurity (RR 2.71, 95% CI 0.75 to 9.75; participants = 127).

Hyperglycaemia treated with insulin (Analysis 3.13): Tan 2008 reported an increase in hyperglycaemia treated with insulin (RR 1.69, 95% CI 1.12 to 2.53; participants = 114).

Cholestasis (Analysis 3.14): Data show no difference in cholestasis (typical RR 1.21, 95% CI 0.76 to 1.94; participants = 241; studies = 2).

Primary outcomes

Mortality to discharge (Analysis 4.1): Data show no difference in mortality (typical RR 0.97, 95% CI 0.66 to 1.42; participants = 567; studies = 5) and no subgroup difference by commencement intake; testing for subgroup differences: P = 0.32, I² = 13.8%.

Neurodevelopmental disability and postnatal growth failure: Trials provided no data.

Secondary outcomes

Days to regain birth weight (Analysis 4.2): Data show a reduction in days to regain birth weight (MD ‐1.86, 95% CI ‐2.79 to ‐0.93; participants = 496; studies = 5) and no subgroup difference by commencement intake; testing for subgroup differences: P = 0.80, I² = 0%.

Maximal weight loss in grams (Analysis 4.3): Studies commencing on high amino acid intake (> 2 to ≤ 3 g/kg/d) reported a reduction in maximal weight loss in grams (MD ‐29.79, 95% CI ‐41.58 to ‐17.99; participants = 185; studies = 2).

Maximal weight loss per cent (Analysis 4.4): Data show no difference in maximal weight loss per cent (MD 0.22, 95% CI ‐1.20 to 1.64; participants = 229; studies = 2) and no subgroup difference by commencement intake; testing for subgroup differences: P = 0.73, I² = 0%.

Weight gain to 1 month (Analysis 4.5): Studies commencing on low amino acid intake (> 1 to ≤ 2 g/kg/d) reported no difference in weight gain to one month (MD 1.48 g/kg/d, 95% CI ‐0.29 to 3.25; participants = 154; studies = 2).

Weight gain to discharge (Analysis 4.6): Burattini 2013 commencing on high amino acid intake (> 2 to ≤ 3 g/kg/d) reported no difference in weight gain to discharge (MD 0.60 g/kg/d, 95% CI ‐0.34 to 1.54; participants = 114).

Linear growth to one month (Analysis 4.7): Clark 2007 commencing on low amino acid intake (> 1 to ≤ 2 g/kg/d) reported no difference in linear growth to one month (MD 0.00 cm/week, 95% CI ‐0.15 to 0.15; participants = 122).

Head circumference to one month (Analysis 4.8): Clark 2007 commencing on low amino acid intake (> 1 to ≤ 2 g/kg/d) reported no difference in head circumference growth to one month (MD 0.00 cm/week, 95% CI ‐0.12 to 0.12; participants = 122).

Head circumference change to discharge (Analysis 4.9): Studies commencing on high amino acid intake (> 2 to ≤ 3 g/kg/d) reported an increase in head circumference change to discharge (MD 0.11 cm/week, 95% CI 0.07 to 0.15; participants = 182; studies = 2).

Days to full enteral feeds (Analysis 4.10): Data show a reduction in days to full enteral feeds with a significant subgroup effect by commencement intake (MD ‐1.08, 95% CI ‐2.42 to 0.25; participants = 431; studies = 5) and no significant subgroup difference by commencement intake; testing for subgroup differences: P = 0.01, I² = 76.9%. For subgroups, studies commencing with low amino acid intake (> 1 to ≤ 2 g/kg/d) reported no difference in days to full enteral feeds (MD 2.47, 95% CI ‐1.73 to 6.68; participants = 147; studies = 2). Studies commencing on high amino acid intake (> 2 to ≤ 3 g/kg/d) reported a reduction in days to full enteral feeds (MD ‐3.32, 95% CI ‐5.39 to ‐1.25; participants = 158; studies = 2). Uthaya 2016 commenced on very high amino acid intake (> 3 g/kg/d) and reported no difference in days to full enteral feeds (MD 0.09, 95% CI ‐1.83 to 2.01; participants = 126).

Late onset sepsis (Analysis 4.11): Data show no difference in late‐onset sepsis (typical RR 0.96, 95% CI 0.72 to 1.29; participants = 772; studies = 8) and no subgroup difference by commencement intake; testing for subgroup differences: P = 0.70, I² = 0%.

Necrotising enterocolitis (Analysis 4.12): Data show no difference in necrotising enterocolitis (typical RR 1.14, 95% CI 0.63 to 2.07; participants = 683; studies = 6) and no subgroup difference by commencement intake; testing for subgroup differences: P = 0.43, I² = 0%.

Chronic lung disease (Analysis 4.13): Data show no difference in chronic lung disease (typical RR 0.81, 95% CI 0.55 to 1.19; participants = 376; studies = 4) and no subgroup difference by commencement intake; testing for subgroup differences: P = 0.13, I² = 58%.

Patent ductus arteriosus (Analysis 4.14): Data show no differences in patent ductus arteriosus (typical RR 0.81, 95% CI 0.60 to 1.10; participants = 236; studies = 2) and no subgroup difference by commencement intake; testing for subgroup differences: P = 0.22, I² = 32%.

Intraventricular haemorrhage (Analysis 4.15): Data show no differences in intraventricular haemorrhage (typical RR 1.09, 95% CI 0.70 to 1.70; participants = 218; studies = 2) and no subgroup difference by commencement intake; testing for subgroup differences: P = 0.79, I² = 0%.

Severe intraventricular haemorrhage (Analysis 4.16): Data show no differences in severe intraventricular haemorrhage (typical RR 0.96, 95% CI 0.46 to 2.02; participants = 402; studies = 4) and no subgroup difference by commencement intake; testing for subgroup differences: P = 0.58, I² = 0%.

Periventricular leukomalacia (Analysis 4.17): Data show a reduction in periventricular leukomalacia (typical RR 0.32, 95% CI 0.10 to 1.00; participants = 447; studies = 4) and no significant subgroup difference of borderline significance by commencement intake; testing for subgroup differences: P = 0.09, I² = 66.1%. For subgroups, data show a reduction in periventricular leukomalacia for infants commenced on low amino acid intake (1.5 to 2 g/kg/d) (typical RR 0.14, 95% CI 0.03 to 0.79; participants = 237; studies = 2) and no reduction in infants commenced on high amino acid intake (2.4 to 2.5 g/kg/d) (typical RR 1.37, 95% CI 0.20 to 9.33; participants = 210; studies = 2).

Severe retinopathy of prematurity (Analysis 4.18): Data show no difference in severe retinopathy of prematurity in infants commencing on low amino acid intake (2 g/kg/d) (typical RR 1.24, 95% CI 0.49 to 3.09; participants = 166; studies = 2). Burattini 2013 commencing with high amino acid intake (2.5 g/kg/d) reported no cases of severe retinopathy of prematurity (114 infants).

Cerebral palsy (Analysis 4.19): A single study reported no difference in cerebral palsy (RR 3.00, 95% CI 0.35 to 25.87; participants = 32) (Blanco 2008).

Developmental delay (Analysis 4.20): A single study reported no difference in developmental delay (RR 3.25, 95% CI 0.35 to 30.19; participants = 100) (Burattini 2013).

Blindness (Analysis 4.21): A single study reported no difference in blindness (RR 2.00, 95% CI 0.20 to 19.91; participants = 32) (Blanco 2008).

Abnormal serum ammonia (Analysis 4.22): A single study reported an abnormal serum ammonia > 122 μmol/L in one infant commencing on low amino acid intake (2 g/kg/d) (Blanco 2008).

Abnormal blood urea nitrogen (Analysis 4.23): Data show an increase in abnormal blood urea nitrogen (various criteria) (typical RR 3.19, 95% CI 2.24 to 4.53; participants = 550; studies = 5) and no subgroup difference by commencement intake; testing for subgroup differences: P = 0.25, I² = 27.2% (Vaidya 1995). For subgroups, a single study commencing with very low amino acid intake (≤ 1 g/kg/d) reported an increase in abnormal blood urea nitrogen (RR 10.74, 95% CI 1.45 to 79.59; participants = 85). For studies commencing low amino acid intake (> 1 to ≤ 2 g/kg/d), data show an increase in abnormal blood urea nitrogen (typical RR 12.29, 95% CI 1.66 to 90.79; participants = 183; studies = 2). A single study commencing on high amino acid intake (> 2 to ≤ 3 g/kg/d) reported an increase in abnormal blood urea nitrogen (RR 2.44, 95% CI 1.47 to 4.05; participants = 114) (Burattini 2013). Another single study commencing on very high amino acid intake (> 3 g/kg/d) reported an increase in abnormal blood urea nitrogen (RR 2.73, 95% CI 1.64 to 4.54; participants = 168) (Uthaya 2016).

Hyperglycaemia (Analysis 4.24): Data show a reduction in hyperglycaemia (typical RR 0.54, 95% CI 0.36 to 0.82; participants = 463; studies = 4) and no subgroup difference by commencement intake; testing for subgroup differences: P = 0.55, I² = 0%. For subgroups, a single study commencing infants on very low amino acid intake (0.5 g/kg/d) reported no difference in hyperglycaemia (RR 1.95, 95% CI 0.18 to 20.74; participants = 85) (Vaidya 1995). Data show a reduction in hyperglycaemia in studies commencing infants on high amino acid intake (2.4 to 2.5 g/kg/d) (typical RR 0.51, 95% CI 0.30 to 0.87; participants = 210; studies = 2) and no difference in hyperglycaemia in studies commencing infants on very high amino acid intake (> 3 g/kg/d) (RR 0.53, 95% CI 0.26 to 1.06; participants = 168).

Hyperglycaemia treated with insulin (Analysis 4.25): Data show no difference in hyperglycaemia treated with insulin (typical RR 0.62, 95% CI 0.35 to 1.08; participants = 282; studies = 2) and no subgroup difference by commencement intake; testing for subgroup differences: P = 0.59, I² = 0%.

Hypoglycaemia (Analysis 4.26): Data show no difference in hypoglycaemia (typical RR 1.03, 95% CI 0.70 to 1.50; participants = 253; studies = 2) and no subgroup difference by commencement intake; testing for subgroup differences: P = 0.76, I² = 0%.

Metabolic acidosis (Analysis 4.27): Data show no difference in metabolic acidosis (typical RR 2.05, 95% CI 0.94 to 4.47; participants = 253; studies = 2) and no subgroup difference by commencement intake; testing for subgroup differences: P = 0.27, I² = 17.4%.

Cholestasis (Analysis 4.28): Data show no difference in cholestasis (typical RR 1.34, 95% CI 0.71 to 2.50; participants = 375; studies = 3) and no subgroup difference by commencement intake; testing for subgroup differences: P = 0.13, I² = 50.4%.

Primary outcomes

Mortality to discharge (Analysis 5.1): Data show no difference in mortality (typical RR 0.97, 95% CI 0.66 to 1.42; participants = 567; studies = 5) and no subgroup difference according to maximal intake; testing for subgroup differences: P = 0.40, I² = 0%.

Neurodevelopmental disability or postnatal growth failure: Trials provided no data.

Secondary outcomes

Days to regain birth weight (Analysis 5.2): Data show a reduction in days to regain birth weight (MD ‐1.86, 95% CI ‐2.79 to ‐0.93; participants = 496; studies = 5) and no subgroup difference according to maximal intake; testing for subgroup differences: P = 0.67, I² = 0%. For subgroups, Vaidya 1995 compared low maximal amino acid intake (3 g/kg/d) versus no amino acid intake and reported no difference in days to regain birth weight (MD ‐1.00, 95% CI ‐5.03 to 3.03; participants = 85). Studies of infants receiving high maximal amino acid intake (> 3 to ≤ 4 g/kg/d) reported a reduction in days to regain birth weight (MD ‐1.91, 95% CI ‐2.87 to ‐0.95; participants = 411; studies = 4).

Maximal weight loss in gram (Analysis 5.3): Studies of high maximal amino acid intake (> 3 to ≤ 4 g/kg/d) reported a reduction in maximal weight loss in grams (MD ‐29.79, 95% CI ‐41.58 to ‐17.99; participants = 185; studies = 2).

Maximal weight loss per cent (Analysis 5.4): Studies of infants receiving high amino acid intake (> 3 to ≤ 4 g/kg/d) reported no difference in maximal weight loss per cent (MD 0.22, 95% CI ‐1.20 to 1.64; participants = 229; studies = 2).

Weight gain to one month (Analysis 5.5): Studies of infants receiving high amino acid intake (> 3 to ≤ 4 g/kg/d) reported no difference in weight gain to one month (MD 1.48 g/kg/d, 95% CI ‐0.29 to 3.25; participants = 154; studies = 2).

Weight gain to discharge (Analysis 5.6): A single study reported no difference in weight gain to discharge (MD 0.60 g/kg/d, 95% CI ‐0.34 to 1.54; participants = 114) (Burattini 2013).

Linear growth to one month (Analysis 5.7): A single study reported no difference in weight gain to discharge (MD 0.00 cm/week, 95% CI ‐0.15 to 0.15; participants = 122) (Clark 2007).

Head circumference growth to one month (Analysis 5.8): A single study reported no difference in weight gain to discharge (MD 0.00, 95% CI ‐0.12 to 0.12; participants = 122) (Clark 2007).

Head circumference growth to discharge (Analysis 5.9): Studies of high maximal amino acid intake (> 3 to ≤ 4 g/kg/d) reported an increase in head circumference growth to discharge (MD 0.11, 95% CI 0.07 to 0.15; participants = 182; studies = 2).

Days to full enteral feeds (Analysis 5.10): Studies of high maximal amino acid intake (> 3 to ≤ 4 g/kg/d) reported a reduction in days to full enteral feeds (MD ‐1.08, 95% CI ‐2.42 to 0.25; participants = 431; studies = 5).

Late‐onset sepsis (Analysis 5.11): Data show no difference in late‐onset sepsis (typical RR 0.96, 95% CI 0.72 to 1.29; participants = 772; studies = 8) and no subgroup difference according to maximal intake; testing for subgroup differences: P = 0.55, I² = 0%.

Necrostising enterocolitis (Analysis 5.12): Studies of high maximal amino acid intake (> 3 to ≤ 4 g/kg/d) reported no difference in necrotising enterocolitis (typical RR 1.14, 95% CI 0.63 to 2.07; participants = 683; studies = 6).

Chronic lung disease (Analysis 5.13): Studies of high maximal amino acid intake (> 3 to ≤ 4 g/kg/d) reported no difference in chronic lung disease (typical RR 0.81, 95% CI 0.55 to 1.19; participants = 376; studies = 4).

Patent ductus arteriosus (Analysis 5.14): Studies of high maximal amino acid intake (> 3 to ≤ 4 g/kg/d) reported no difference in patent ductus arteriosus (typical RR 0.81, 95% CI 0.60 to 1.10; participants = 236; studies = 2).

Intraventricular haemorrhage (Analysis 5.15): Studies of high maximal amino acid intake (> 3 to ≤ 4 g/kg/d) reported no difference in intraventricular haemorrhage (typical RR 1.09, 95% CI 0.70 to 1.70; participants = 218; studies = 2).

Severe intraventricular haemorrhage (Analysis 5.16): Studies of high maximal amino acid intake (> 3 to ≤ 4 g/kg/d) reported no difference in severe intraventricular haemorrhage (typical RR 0.96, 95% CI 0.46 to 2.02; participants = 402; studies = 4).

Periventricular leukomalacia (Analysis 5.17): Studies of high maximal amino acid intake (> 3 to ≤ 4 g/kg/d) reported a reduction in periventricular leukomalacia (typical RR 0.32, 95% CI 0.10 to 1.00; RD ‐0.04, 95% CI ‐0.07, ‐0.00; P = 0.05; participants = 447; studies = 4).

Severe retinopathy of prematurity (Analysis 5.18): Studies of high maximal amino acid intake (> 3 to ≤ 4 g/kg/d) reported no difference in severe retinopathy of prematurity (typical RR 1.24, 95% CI 0.49 to 3.09; participants = 280; studies = 3).

Cerebral palsy (Analysis 5.19): In infants receiving high maximal amino acid intake (4 g/kg/d), Blanco 2008 reported no difference in cerebral palsy (RR 3.00, 95% CI 0.35 to 25.87; participants = 32).

Developmental delay (Analysis 5.20): In infants receiving high maximal amino acid intake (4 g/kg/d), Burattini 2013 reported no difference in developmental delay (RR 3.25, 95% CI 0.35 to 30.19; participants = 100).

Blindness (Analysis 5.21): In infants receiving high maximal amino acid intake (4 g/kg/d), Blanco 2008 reported no difference in blindness (RR 2.00, 95% CI 0.20 to 19.91; participants = 32).

Abnormal serum ammonia (Analysis 5.22): Blanco 2008 reported an abnormal serum ammonia > 122 μmol/L in one infant receiving high maximal amino acid intake (4 g/kg/d).

Abnormal blood urea nitrogen (Analysis 5.23): Data show an increase in abnormal blood urea nitrogen (various criteria) (typical RR 3.19, 95% CI 2.24 to 4.53; participants = 550; studies = 5) and no subgroup difference according to maximal intake; testing for subgroup differences: P = 0.21, I² = 36.2%. For subgroups, a single study of low maximal amino acid intake (3.0 g/kg/d) reported an increase in abnormal blood urea nitrogen (criteria not reported) (RR 10.74, 95% CI 1.45 to 79.59; participants = 85) (Vaidya 1995). Studies of high maximal amino acid intake (3.5 to 4 g/kg/d) reported an increase in abnormal blood urea nitrogen (typical RR 2.93, 95% CI 2.05 to 4.18; participants = 465; studies = 4).

Hyperglycaemia (Analysis 5.24): Data show a reduction in hyperglycaemia (typical RR 0.54, 95% CI 0.36 to 0.82; participants = 463; studies = 4) and no subgroup difference according to maximal intake; testing for subgroup differences: P = 0.28, I² = 15.8%. For subgroups, a single study of low maximal amino acid intake (3.0 g/kg/d) reported no difference in hyperglycaemia (85 infants, RR 1.95, 95% CI 0.18, 20.74) (Vaidya 1995). Studies of infants on high maximal amino acid intake (> 3 to ≤ 4 g/kg/d) reported a reduction in hyperglycaemia (typical RR 0.51, 95% CI 0.34 to 0.79; participants = 378; studies = 3).

Hyperglycaemia treated with insulin (Analysis 5.25): Data show no difference in hyperglycaemia treated with insulin (typical RR 0.62, 95% CI 0.35 to 1.08; participants = 282; studies = 2) and no subgroup difference according to maximal intake; testing for subgroup differences: P = 0.59, I² = 0%.

Hypoglycaemia (Analysis 5.26): Data show no difference in hypoglycaemia (typical RR 1.03, 95% CI 0.70 to 1.50; participants = 253; studies = 2) and no subgroup difference according to maximal intake; testing for subgroup differences: P = 0.76, I² = 0%.

Metabolic acidosis (Analysis 5.27): Data show no difference in metabolic acidosis (typical RR 2.05, 95% CI 0.94 to 4.47; participants = 253; studies = 2) and no subgroup difference according to maximal intake; testing for subgroup differences: P = 0.27, I² = 17.4%.

Cholestasis (Analysis 5.28): Data show no difference in cholestasis (typical RR 1.34, 95% CI 0.71 to 2.50; participants = 375; studies = 3) and no subgroup difference according to maximal intake; testing for subgroup differences: P = 0.11, I² = 66.6%.

Primary outcomes

Mortality to discharge (Analysis 6.1): Data show no difference in mortality (typical RR 0.90, 95% CI 0.69 to 1.17; participants = 1407; studies = 14) and no significant subgroup difference according to management of caloric balance; testing for subgroup differences: P = 0.39, I² = 0%.

Neurodevelopmental disability (Analysis 6.2): Studies that increased amino acids and provided isocaloric non‐protein intake reported no difference in neurodevelopmental disability (typical RR 1.04, 95% CI 0.48 to 2.23; participants = 201; studies = 2).

Postnatal growth failure at discharge (Analysis 6.3): Data show a reduction in postnatal growth failure at discharge (typical RR 0.74, 95% CI 0.56 to 0.97; participants = 203; studies = 3) and no significant subgroup difference according to management of caloric balance; testing for subgroup differences: P = 0.93, I² = 0%. Studies that increase amino acids and non‐protein caloric intake reported a reduction in postnatal growth failure at discharge (typical RR 0.73, 95% CI 0.55 to 0.98; participants = 92; studies = 2).

Secondary outcomes

Days to regain birth weight (Analysis 6.4): Data show a reduction in days to regain birth weight overall (MD ‐1.14, 95% CI ‐1.73 to ‐0.56; participants = 950; studies = 13) with no significant subgroup difference according to management of caloric balance; testing for subgroup differences: P = 0.68, I² = 0%. Studies that increased amino acids and provided isocaloric non‐protein intake reported a reduction in days to regain birth weight (MD ‐1.06, 95% CI ‐1.77 to ‐0.34; participants = 615; studies = 8). Studies that increased amino acids and non‐protein caloric intake reported a reduction in days to regain birth weight (MD ‐1.32, 95% CI ‐2.33 to ‐0.31; participants = 335; studies = 5).

Maximal weight loss grams (Analysis 6.5): Data show a significant reduction in maximal weight loss overall (MD ‐22.71, 95% CI ‐33.68 to ‐11.74; participants = 235; studies = 3) and a significant subgroup difference according to management of caloric balance; testing for subgroup differences: P = 0.001, I² = 90.2%. Studies that increased amino acids and provided isocaloric non‐protein intake reported a reduction in maximal weight loss (MD ‐29.79, 95% CI ‐41.58 to ‐17.99; participants = 185; studies = 2). A single study that increased amino acids and non‐protein calorie intake reported no differences (50 infants; MD 22.60 g, 95% CI ‐7.25 to 52.45) (Can 2012).

Maximal weight loss per cent (Analysis 6.6): Data show no difference in maximal weight loss per cent overall (MD ‐0.33, 95% CI ‐1.61 to 0.96; participants = 288; studies = 4) and a significant subgroup difference according to management of caloric balance; testing for subgroup differences: P = 0.03, I² = 78.7%. Studies that increased amino acids and provided isocaloric non‐protein intake reported no difference in maximal weight loss per cent (MD 0.25, 95% CI ‐1.13 to 1.64; participants = 246; studies = 3). A single study that increased amino acids and non‐protein caloric intake reported a reduction in maximal weight loss per cent (MD ‐3.80, 95% CI ‐7.20 to ‐0.40; participants = 42) (Pappoe 2009).

Weight gain to one month (Analysis 6.7): Studies that increased amino acids and provided isocaloric non‐protein intake reported a reduction in weight gain (g/kg/d) to one month (MD ‐1.50, 95% CI ‐2.56 to ‐0.44; participants = 373; studies = 4).

Weight gain to discharge (Analysis 6.8): Data show no difference overall in weight gain (g/kg/d) to discharge (MD 0.76, 95% CI ‐0.02 to 1.54; participants = 291; studies = 4) and no significant subgroup difference according to management of caloric balance; testing for subgroup differences: P = 0.72, I² = 0%.

Linear growth to one month (Analysis 6.9): A single study that increased amino acids and provided isocaloric non‐protein intake reported a reduction in linear growth to one month (MD ‐0.00 cm/week, 95% CI ‐0.15, ‐0.15; participants = 122) (Clark 2007).

Linear growth to discharge (Analysis 6.10): Another single study that increased amino acids and provided isocaloric non‐protein intake reported a reduction in linear growth to discharge (MD ‐0.27 cm/week, 95% CI ‐0.40, ‐0.14; participants = 123) (Balasubramanian 2013).

Head circumference growth to one month (Analysis 6.11): Data show no difference overall in head circumference growth up to one month (MD 0.01 cm/week, 95% CI ‐0.05 to 0.07; participants = 380; studies = 3) and a significant subgroup difference according to management of caloric balance; testing for subgroup differences: P < 0.00001, I² = 95%. Studies that increased amino acids and provided isocaloric non‐protein intake reported a decrease in head growth to one month (MD ‐0.16, 95% CI ‐0.25 to ‐0.07; participants = 245; studies = 2). A single study that increased amino acids and non‐protein caloric intake reported an increase in head circumference growth to one month (MD 0.13 cm/week, 95% CI 0.05, 0.20; participants 135) (Morgan 2014).

Head circumference growth to discharge (Analysis 6.12): Studies that increased amino acids and provided isocaloric non‐protein intake reported an increase in head growth to discharge (MD 0.09, 95% CI 0.06 to 0.13; participants = 315; studies = 4).

Weight change z‐score to discharge (Analysis 6.13): Studies that increased amino acids and provided isocaloric non‐protein intake reported no difference in weight change z‐score to discharge (MD 0.01, 95% CI ‐0.33 to 0.36; participants = 207; studies = 2).

Head circumference change z‐score to one month (Analysis 6.14): Data show an increase in head circumference change z‐score to one month (MD 0.27, 95% CI 0.08 to 0.46; participants = 231; studies = 2) and subgroup differences according to management of caloric balance that was borderline significant; testing for subgroup differences: P = 0.09, I² = 66%. For subgroups, a single study that increased amino acids and provided isocaloric non‐protein intake reported no difference in head circumference change z‐score to one month (RR 0.00, 95% CI ‐0.36 to 0.36; participants = 96) (Vlaardingerbroek 2013). Another single study that increased amino acids and non‐protein intake reported an increase in head circumference change z‐score to one month (RR 0.37, 95% CI 0.15, 0.59; participants = 135) (Morgan 2014).

Head circumference change z‐score to discharge (Analysis 6.15): Studies that increased amino acids and provided isocaloric non‐protein intake found no difference in head circumference change z‐score to discharge (MD 0.18, 95% CI ‐0.15 to 0.50; participants = 207; studies = 2).

Days to full enteral feeds (Analysis 6.16): Data show no difference in days to full enteral feeds (MD ‐0.19, 95% CI ‐1.07 to 0.70; participants = 778; studies = 11) and no significant subgroup difference according to management of caloric balance; testing for subgroup differences: P = 0.11, I² = 60.9%. Studies that increased amino acids and provided isocaloric non‐protein intake reported no difference in days to full enteral feeds (MD ‐0.90, 95% CI ‐2.14 to 0.35; participants = 495; studies = 7). Studies that increased amino acids and non‐protein caloric intake reported no difference in days to full enteral feeds (MD 0.56, 95% CI ‐0.71 to 1.83; participants = 283; studies = 4).

Late‐onset sepsis (Analysis 6.17): Data show no difference in late‐onset sepsis (typical RR 0.96, 95% CI 0.79 to 1.18; participants = 1255; studies = 15) and no significant subgroup difference according to management of caloric balance; testing for subgroup differences: P = 0.66, I² = 0%.

Necrotising enterocolitis (Analysis 6.18): Data show no difference in necrotising enterocolitis (typical RR 1.00, 95% CI 0.68 to 1.47; participants = 1301; studies = 14) and no significant subgroup difference according to management of caloric balance; testing for subgroup differences: P = 0.44, I² = 0%.

Chronic lung disease (Analysis 6.19): Data show no difference in chronic lung disease (typical RR 1.04, 95% CI 0.89 to 1.23; participants = 819; studies = 10) and no significant subgroup difference according to management of caloric balance; testing for subgroup differences: P = 0.43, I² = 0%.

Intraventricular haemorrhage (Analysis 6.20): Data show no difference in intraventricular haemorrhage (typical RR 1.08, 95% CI 0.73 to 1.59; participants = 370; studies = 4) and no significant subgroup difference according to management of caloric balance; testing for subgroup differences: P = 0.66, I² = 0%.

Severe intraventricular haemorrhage (Analysis 6.21): Data show no difference in severe intraventricular haemorrhage (typical RR 1.16, 95% CI 0.74 to 1.82; participants = 904; studies = 11) and no significant subgroup difference according to management of caloric balance; testing for subgroup differences: P = 0.44, I² = %.

Periventricular leukomalacia (Analysis 6.22): Data show no difference in periventricular leukomalacia (typical RR 0.55, 95% CI 0.24 to 1.25; participants = 720; studies = 7) and no significant subgroup difference according to management of caloric balance; testing for subgroup differences: P = 0.12, I² = 58%. For subgroups, data show a reduction in periventricular leukomalacia in studies that increased amino acids and provided isocaloric non‐protein intake of borderline significance (typical RR 0.32, 95% CI 0.10 to 1.00; participants = 543; studies = 5).

Retinopathy of prematurity (Analysis 6.23): Data show a reduction in retinopathy of prematurity (typical RR 0.44, 95% CI 0.21 to 0.93; participants = 269; studies = 4) and no significant subgroup difference according to management of caloric balance; testing for subgroup differences: P = 0.11, I² = 61%. The single study that increased amino acids and provided isocaloric non‐protein intake reported no difference in retinopathy of prematurity (123 infants; RR 1.58, 95% CI 0.27 to 9.10). Studies that increased amino acids and non‐protein caloric intake reported a reduction in retinopathy of prematurity (typical RR 0.32, 95% CI 0.13 to 0.77; participants = 146; studies = 3).

Severe retinopathy of prematurity (Analysis 6.24): Data show no difference in severe retinopathy of prematurity (typical RR 0.96, 95% CI 0.56 to 1.63; participants = 672; studies = 8) and no significant subgroup difference according to management of caloric balance; testing for subgroup differences: P = 0.20, I² = 38%.

Cerebral palsy (Analysis 6.25): Data show no difference in cerebral palsy (typical RR 4.00, 95% CI 0.89 to 17.97; participants = 122; studies = 2) for studies that increased amino acids and provided isocaloric non‐protein intake

Developmental delay at ≥ 18 months (Analysis 6.26): Studies that increased amino acids and provided isocaloric non‐protein intake reported no difference in developmental delay at ≥ 18 months (typical RR 1.35, 95% CI 0.52 to 3.53; participants = 301; studies = 3).

Blindness (Analysis 6.27): Data show no difference in blindness (typical RR 2.00, 95% CI 0.20 to 19.91; participants = 122; studies = 2) for studies that increased amino acids and provided isocaloric non‐protein intake.

Deafness (Analysis 6.28): A single study that increased amino acids and provided isocaloric non‐protein intake reported that no infants were deaf (Vlaardingerbroek 2013).

Abnormal serum ammonia (Analysis 6.29): A single study that increased amino acids and provided isocaloric non‐protein intake reported an abnormal serum ammonia > 122 μmol/L in one infant on higher amino acid intake (Blanco 2008).

Abnormal blood urea nitrogen (Analysis 6.30): Data show an increase in abnormal blood urea nitrogen overall (typical RR 2.77, 95% CI 2.13 to 3.61; participants = 688; studies = 7) with no significant subgroup differences according to management of caloric balance; testing for subgroup differences: P = 0.22, I² = 33.3%. Data show a significant increase in abnormal blood urea nitrogen in studies that increased amino acids and provided isocaloric non‐protein intake (typical RR 2.60, 95% CI 2.00 to 3.40; participants = 561; studies = 5) and in studies that increased both amino acids and non‐protein calorie intake (typical RR 6.45, 95% CI 1.55 to 26.84; participants = 127; studies = 2).

Hyperglycaemia (Analysis 6.31): Data show no difference in hyperglycaemia overall (typical RR 0.51, 95% CI 0.34 to 0.79; participants = 378; studies = 3) and a significant subgroup difference according to management of caloric balance; testing for subgroup differences: P = 0.002, I² = 89.3%. For subgroups, studies that increased amino acids and provided isocaloric non‐protein intake reported a reduction in hyperglycaemia (typical RR 0.51, 95% CI 0.34 to 0.79; participants = 378; studies = 3). Studies that increased amino acids and non‐protein calorie nutrition reported no difference (typical RR 1.51, 95% CI 0.88 to 2.62; participants = 127; studies = 2).

Hyperglycaemia treated with insulin (Analysis 6.32): Data show an increase in hyperglycaemia treated with insulin overall (typical RR 1.24, 95% CI 0.93 to 1.66; participants = 534; studies = 5) and a significant subgroup difference according to management of caloric balance; tests for subgroup differences: P = 0.001, I² = 90.1%. For subgroups, studies that increased amino acids and provided isocaloric non‐protein intake reported no difference in hyperglycaemia treated with insulin (typical RR 0.76, 95% CI 0.49 to 1.19; participants = 378; studies = 3). Studies that increased amino acids and non‐protein calorie nutrition reported an increase (typical RR 2.00, 95% CI 1.35 to 2.98; participants = 156; studies = 2).

Hypoglycaemia (Analysis 6.33): Data show no significant effect overall in hypoglycaemia (typical RR 1.17, 95% CI 0.84 to 1.63; participants = 376; studies = 3) and no significant subgroup difference according to management of caloric balance; testing for subgroup differences: P = 0.94, I² = 0%.

Metabolic acidosis (Analysis 6.34): Data show no significant effect overall in metabolic acidosis (typical RR 2.05, 95% CI 0.94 to 4.47; participants = 305; studies = 4) and no significant subgroup difference according to management of caloric balance; testing for subgroup differences: P = 0.27, I² = 17.4%.

Cholestasis (Analysis 6.35): Data show no difference in cholestasis (typical RR 1.26, 95% CI 0.86 to 1.84; participants = 616; studies = 5) and no significant subgroup difference according to management of caloric balance; testing for subgroup differences: P = 0.33, I² = 0%.

Primary outcomes

Mortality to discharge (Analysis 7.1): Data show no difference in mortality (typical RR 0.90, 95% CI 0.69 to 1.17; participants = 1407; studies = 14).

Neurodevelopmental disability at two years (Analysis 7.2): Data show no difference in neurodevelopmental disability at two years (typical RR 1.04, 95% CI 0.48 to 2.23; participants = 201; studies = 2).

Postnatal growth failure (Analysis 7.3): Data show a reduction in postnatal growth failure at discharge (typical RR 0.74, 95% CI 0.56 to 0.97; participants = 203; studies = 3). A single study reported no difference in growth failure at two years (RR 0.66, 95% CI 0.33 to 1.32; participants = 111) (te Braake 2005).

Secondary outcomes

Days to regain birth weight (Analysis 7.4): Data show a significant reduction in days to regain birth weight (MD ‐0.78, 95% CI ‐1.46 to ‐0.11; participants = 800; studies = 10).

Maximal weight loss in grams (Analysis 7.5): Data show a reduction in maximal weight loss in grams (MD ‐20.37, 95% CI ‐32.68 to ‐8.05; participants = 139; studies = 2).

Maximal weight loss per cent (Analysis 7.6): Data show no difference in maximal weight loss per cent (MD ‐0.38, 95% CI ‐1.69 to 0.93; participants = 271; studies = 3).

Weight gain (Analysis 7.7): Data show a reduction in weight gain to one month (MD ‐1.50, 95% CI ‐2.56 to ‐0.44; participants = 373; studies = 4) and no difference in weight gain to discharge (MD 0.72, 95% CI ‐0.09 to 1.52; participants = 254; studies = 3).

Linear growth (Analysis 7.8): Data show a reduction in linear growth to one month (MD ‐0.16, 95% CI ‐0.26 to ‐0.06; participants = 245; studies = 2).

Head circumference growth (Analysis 7.9): Data show no difference in head circumference growth to one month (MD 0.01, 95% CI ‐0.04 to 0.06; participants = 476; studies = 4) and an increase in head circumference growth to discharge (MD 0.08, 95% CI 0.05 to 0.12; participants = 182; studies = 2).

Weight change in z‐score (Analysis 7.10): A single study reported no difference in weight change in z‐score to one month (MD ‐0.20, 95% CI ‐0.62 to 0.22; participants = 96) (Vlaardingerbroek 2013). Data show no difference in weight change in z‐score to discharge (MD 0.01, 95% CI ‐0.33 to 0.36; participants = 207; studies = 2) and no difference in weight change in z‐score post discharge (MD 0.13, 95% CI ‐0.26 to 0.52; participants = 201; studies = 2).

Head circumference change in z‐score (Analysis 7.11): Data show an increase in head circumference change in z‐score to one month (MD 0.27, 95% CI 0.08 to 0.46; participants = 231; studies = 2). Meta‐analysis revealed no difference in head circumference change in z‐score to discharge (MD 0.18, 95% CI ‐0.15 to 0.50; participants = 207; studies = 2) and no difference in head circumference change in z‐score post discharge (MD 0.25, 95% CI ‐0.14 to 0.64; participants = 201; studies = 2).

Days to full enteral feeds (Analysis 7.12): Data show no difference in days to full enteral feeds (MD ‐0.19, 95% CI ‐1.07 to 0.70; participants = 778; studies = 11).

Late‐onset sepsis (Analysis 7.13): Data show no difference in late‐onset sepsis (typical RR 0.96, 95% CI 0.79 to 1.18; participants = 1255; studies = 15).

Necrotising enterocolitis (Analysis 7.14): Data show no difference in necrotising enterocolitis (typical RR 1.00, 95% CI 0.68 to 1.47; participants = 1301; studies = 14).

Chronic lung disease (Analysis 7.15): Data show no difference in chronic lung disease (typical RR 1.04, 95% CI 0.89 to 1.23; participants = 819; studies = 10).

Intraventricular haemorrhage (Analysis 7.16): Data show no difference in intraventricular haemorrhage (typical RR 1.12, 95% CI 0.74 to 1.69; participants = 341; studies = 3).

Severe intraventricular haemorrhage (Analysis 7.17): Data show no difference in severe intraventricular haemorrhage (typical RR 1.16, 95% CI 0.74 to 1.82; participants = 904; studies = 11).

Periventricular leukomalacia (Analysis 7.18): Data show no difference in periventricular leukomalacia (typical RR 0.48, 95% CI 0.20 to 1.17; participants = 624; studies = 6).

Retinopathy of prematurity (Analysis 7.19): Data show a reduction in retinopathy of prematurity (typical RR 0.44, 95% CI 0.21 to 0.93; participants = 269; studies = 4).

Severe retinopathy of prematurity (Analysis 7.20): Data show no difference in severe retinopathy of prematurity (typical RR 0.96, 95% CI 0.56 to 1.63; participants = 672; studies = 8).

Cerebral palsy (Analysis 7.21): Data show no difference in cerebral palsy (typical RR 4.00, 95% CI 0.89 to 17.97; participants = 122; studies = 2).

Developmental delay (Analysis 7.22): Data show no difference in developmental delay (typical RR 1.35, 95% CI 0.52 to 3.53; participants = 301; studies = 3).

Blindness (Analysis 7.23): Data show no difference in blindness (typical RR 2.00, 95% CI 0.20 to 19.91; participants = 122; studies = 2).

Deafness (Analysis 7.24): A single study reported no difference in infants with deafness in either group (Vlaardingerbroek 2013).

Abnormal serum ammonia (Analysis 7.25): A single study reported no difference in abnormal serum ammonia >122 μmol/L (RR 3.10, 95% CI 0.13 to 73.16; participants = 61) (Blanco 2008).

Abnormal blood urea nitrogen (Analysis 7.26): Data show an increase in abnormal blood urea nitrogen (typical RR 2.77, 95% CI 2.13 to 3.61; participants = 688; studies = 7).

Hyperglycaemia (Analysis 7.27): Data show a reduction in hyperglycaemia (typical RR 0.66, 95% CI 0.45 to 0.96; participants = 409; studies = 4).

Hyperglycaemia treated with insulin (Analysis 7.28): Data show no difference in hyperglycaemia treated with insulin (typical RR 1.24, 95% CI 0.93 to 1.66; participants = 534; studies = 5).

Hypoglycaemia (Analysis 7.29): Data show no difference in hypoglycaemia (typical RR 1.17, 95% CI 0.84 to 1.63; participants = 376; studies = 3).

Metabolic acidosis (Analysis 7.30): Data show no difference in metabolic acidosis (typical RR 2.05, 95% CI 0.94 to 4.47; participants = 253; studies = 2).

Cholestasis (Analysis 7.31): Data show no difference in cholestasis (typical RR 1.26, 95% CI 0.86 to 1.84; participants = 616; studies = 5).

Primary outcomes

Mortality to discharge (Analysis 8.1): Data show no difference in mortality (typical RR 0.90, 95% CI 0.69 to 1.17; participants = 1407; studies = 14) and no significant subgroup difference according to age at commencement of amino acid; testing for subgroup differences: P = 0.45, I² = 0%.

Neurodevelopmental disability (Analysis 8.2): Studies that commenced amino acids at < 24 hours' age reported no difference in neurodevelopmental disability (typical RR 1.04, 95% CI 0.48 to 2.23; participants = 201; studies = 2).

Postnatal growth failure at discharge (Analysis 8.3): Studies that commenced amino acids at < 24 hours' age reported a significant reduction in postnatal growth failure at discharge (typical RR 0.74, 95% CI 0.56 to 0.97; participants = 203; studies = 3).

Secondary outcomes

Days to regain birth weight (Analysis 8.4): Data show a reduction in days to regain birth weight overall (MD ‐1.14, 95% CI ‐1.73 to ‐0.56; participants = 950; studies = 13) and no significant subgroup difference according to age at commencement of amino acid; testing for subgroup differences: P = 0.94, I² = 0%. Data show a reduction in days to regain birth weight in studies that commenced higher amino acid intakes at < 24 hours (MD ‐1.15, 95% CI ‐1.74 to ‐0.56; participants = 865; studies = 12). A single study that commenced higher amino acid intakes at ≥ 24 to 48 hours age reported no difference (MD ‐1.00, 95% CI ‐5.03 to 3.03; participants = 85) (Vaidya 1995).

Weight loss in grams (Analysis 8.5): Data show a reduction in weight loss in grams in studies that commenced higher amino acids at < 24 hours (MD ‐22.71, 95% CI ‐33.68 to ‐11.74; participants = 235; studies = 3).

Weight loss per cent (Analysis 8.6): Data show no difference in weight loss per cent in studies that commenced higher amino acids at < 24 hours (MD ‐0.33, 95% CI ‐1.61 to 0.96; participants = 288; studies = 4).

Weight gain to one month (Analysis 8.7): A single study that commenced higher amino acid intakes at ≥ 24 to 48 hours age reported no difference in weight gain to one month (MD 1.50, 95% CI ‐0.27 to 3.27; participants = 122) (Clark 2007).

Weight gain to discharge (Analysis 8.8): Data show no difference in weight gain (g/kg/d) to discharge overall (MD ‐0.16, 95% CI ‐0.87 to 0.54; participants = 446; studies = 6) and no significant subgroup difference according to age at commencement of amino acid; testing for subgroup differences: P = 0.08, I² = 67.8%.

Linear growth to one month (Analysis 8.9): Data show a reduction in linear growth to one month overall (MD ‐0.16, 95% CI ‐0.26 to ‐0.06; participants = 245; studies = 2) with significant subgroup differences according to age at commencement of amino acid; testing for subgroup differences: P = 0.007, I² = 86.2%. A single study that commenced amino acids at < 24 hours reported a reduction in linear growth to one month (MD ‐0.27, 95% CI ‐0.40 to ‐0.14; participants = 123) (Balasubramanian 2013). Another single study that commenced amino acids at ≥ 24 to 48 hours' age reported no difference in linear growth to one month (MD 0.00, 95% CI ‐0.15 to 0.15; participants = 122) (Clark 2007).

Head circumference growth to one month (Analysis 8.10): Data show no difference in head circumference growth to one month (MD 0.01, 95% CI ‐0.04 to 0.06; participants = 476; studies = 4) and no significant subgroup difference according to age at commencement of amino acid; testing for subgroup differences: P = 1.00, I² = 0. Studies that commenced amino acids at < 24 hours reported no difference in head circumference growth to one month (MD 0.01, 95% CI ‐0.06 to 0.08; participants = 258; studies = 2). Studies that commenced amino acids at ≥ 24 to 48 hours reported no difference in head circumference growth to one month (MD 0.01, 95% CI ‐0.07 to 0.09; participants = 218; studies = 2).

Head circumference growth to discharge (Analysis 8.11): Data show an increase in head circumference growth to discharge (MD 0.09, 95% CI 0.06 to 0.13; participants = 315; studies = 4) and a significant subgroup difference according to age at commencement of amino acid; testing for subgroup differences: P = 0.007, I² = 86.2. Studies that commenced amino acids at < 24 hours reported an increase in head circumference growth to discharge (MD 0.12, 95% CI 0.08 to 0.17; participants = 219; studies = 3). A single study that commenced amino acids at ≥ 24 to 48 hours' age reported no difference (MD 0.03, 95% CI ‐0.03 to 0.09; participants = 96) (Vlaardingerbroek 2013).

Weight gain change in z‐score (Analysis 8.12): Data show no difference in weight gain change in z‐score to discharge (MD 0.01, 95% CI ‐0.33 to 0.36; participants = 207; studies = 2) and no significant subgroup difference according to age at commencement of amino acid; testing for subgroup differences: P = 0.17, I² = 48.1%.

Head circumference change in z‐score to one month (Analysis 8.13): Data show an increase in head circumference change in z‐score up to one month of age (MD 0.27, 95% CI 0.08 to 0.46; participants = 231; studies = 2) and no significant subgroup difference according to age at commencement of amino acid; testing for subgroup differences: P = 0.09, I² = 66%. A single study that commenced amino acids at < 24 hours' age reported an increase in head circumference change z‐score up to one month (MD 0.37, 95% CI 0.15 to 0.59; participants = 135) (Morgan 2014). Another single study that commenced amino acids at ≥ 24 to 48 hours reported no difference (MD 0.00, 95% CI ‐0.36 to 0.36; participants = 96) (Vlaardingerbroek 2013).

Head circumference change in z‐score to discharge (Analysis 8.14): Data show no difference in head circumference change in z‐score to discharge (MD 0.18, 95% CI ‐0.15 to 0.50; participants = 207; studies = 2) and no significant subgroup difference according to age at commencement of amino acid; testing for subgroup differences: P = 0.10, I² = 62.7%.

Days to full enteral feeds (Analysis 8.15): All studies commenced amino acids at < 24 hours. Data show no difference in days to full enteral feeds (MD ‐0.19, 95% CI ‐1.07 to 0.70; participants = 778; studies = 11).

Late‐onset sepsis (Analysis 8.16): Data who no difference in late‐onset sepsis (typical RR 0.96, 95% CI 0.79 to 1.18; participants = 1255; studies = 15) and no significant subgroup difference according to age at commencement of amino acid; testing for subgroup differences: P = 0.72, I² = 0%.

Necrotising enterocolitis (Analysis 8.17): Data show no difference in necrotising enterocolitis (typical RR 1.00, 95% CI 0.68 to 1.47; participants = 1301; studies = 14) and no significant subgroup difference according to age at commencement of amino acid; testing for subgroup differences: P = 0.85, I² = 0%.

Chronic lung disease (Analysis 8.18): Data show no difference in chronic lung disease (typical RR 1.04, 95% CI 0.89 to 1.23; participants = 819; studies = 10) and no significant subgroup difference according to age at commencement of amino acid; testing for subgroup differences: P = 0.23, I² = 29.5%.

Intraventricular haemorrhage (Analysis 8.19): Data show no difference in intraventricular haemorrhage (typical RR 1.12, 95% CI 0.74 to 1.69; participants = 341; studies = 3) and no significant subgroup difference according to age at commencement of amino acid; testing for subgroup differences: P = 0.72, I² = 0%.

Severe intraventricular haemorrhage (Analysis 8.20): Data show no difference in severe intraventricular haemorrhage (typical RR 1.16, 95% CI 0.74 to 1.82; participants = 904; studies = 11) and no significant subgroup difference according to age at commencement of amino acid; testing for subgroup differences: P = 0.97, I² = 0%.

Periventricular leukomalacia (Analysis 8.21): Data show no difference in periventricular leukomalacia (typical RR 0.55, 95% CI 0.24 to 1.25; participants = 720; studies = 7) and no significant subgroup difference according to age at commencement of amino acid; testing for subgroup differences: P = 0.20, I² = 40.3%.

Retinopathy of prematurity (Analysis 8.22): All four studies reporting retinopathy of prematurity commenced amino acids at < 24 hours. Data show a reduction in retinopathy of prematurity (typical RR 0.44, 95% CI 0.21 to 0.93; participants = 269; studies = 4).

Severe retinopathy of prematurity (Analysis 8.23): Data show no difference in severe retinopathy of prematurity (typical RR 0.96, 95% CI 0.56 to 1.63; participants = 672; studies = 8) and no significant subgroup difference according to age at commencement of amino acid; testing for subgroup differences: P = 0.26, I² = 20%.

Cerebral palsy (Analysis 8.24): Studies that commenced amino acids at < 24 hours reported no difference in cerebral palsy (typical RR 4.00, 95% CI 0.89 to 17.97; participants = 122; studies = 2).

Developmental delay at ≥ 18 months (Analysis 8.25): Studies that commenced amino acids at < 24 hours reported no difference in developmental delay at ≥ 18 months (typical RR 1.35, 95% CI 0.52 to 3.53; participants = 301; studies = 3).

Blindness (Analysis 8.26): Studies that commenced amino acids at < 24 hours reported no difference in blindness (typical RR 2.00, 95% CI 0.20 to 19.91; participants = 122; studies = 2).

Deafness (Analysis 8.27): A single study that commenced amino acids at ≥ 24 to < 48 hours reported no infants with deafness in either group (Vlaardingerbroek 2013).

Abnormal serum ammonia (Analysis 8.28): A single study that commenced amino acids at < 24 hours reported an abnormal serum ammonia > 122 μmol/L in one infant on higher amino acid intake (Blanco 2008).

Abnormal blood urea nitrogen (Analysis 8.29): Data show an increase in abnormal blood urea nitrogen overall (typical RR 2.77, 95% CI 2.13 to 3.61; participants = 688; studies = 7) and no significant subgroup difference according to age at commencement of amino acid; testing for subgroup differences: P = 0.23, I² = 31.4%. Data show increases in abnormal blood urea nitrogen in studies commencing amino acid at < 24 hours (typical RR 2.85, 95% CI 2.00 to 4.06; participants = 385; studies = 4); studies commencing amino acid at ≥ 24 to 48 hours (typical RR 2.20, 95% CI 1.50 to 3.23; participants = 218; studies = 2); and one study commencing amino acid at ≥ 48 to 72 hours (RR 10.74, 95% CI 1.45 to 79.59; participants = 85) (Vaidya 1995).

Hyperglycaemia (Analysis 8.30): Data show a reduction in hyperglycaemia overall (typical RR 0.69, 95% CI 0.49 to 0.96; participants = 505; studies = 5) and no significant subgroup difference according to age at commencement of amino acid; testing for subgroup differences: P = 0.38, I² = 0%.

Hyperglycaemia treated with insulin (Analysis 8.31): Data show no difference in hyperglycaemia treated with insulin (typical RR 1.24, 95% CI 0.93 to 1.66; participants = 534; studies = 5) and no significant subgroup difference according to age at commencement of amino acid; testing for subgroup differences: P = 0.82, I² = 0%. Subgroup analysis of studies that commenced amino acids at < 24 hours revealed no difference in hyperglycaemia treated with insulin (typical RR 1.26, 95% CI 0.92 to 1.72; participants = 438; studies = 4). A single study that commenced amino acids at ≥ 24 to < 48 hours reported no difference (RR 1.15, 95% CI 0.54 to 2.45; participants = 96) (Vlaardingerbroek 2013).

Hypoglycaemia (Analysis 8.32): Data show no difference in hypoglycaemia overall (typical RR 1.17, 95% CI 0.84 to 1.63; participants = 376; studies = 3) and no significant subgroup difference according to age at commencement of amino acid; testing for subgroup differences: P = 0.94, I² = 0%.

Metabolic acidosis (Analysis 8.33): Data show no difference in metabolic acidosis overall (typical RR 2.05, 95% CI 0.94 to 4.47; participants = 305; studies = 4) and no significant subgroup difference according to age at commencement of amino acid; testing for subgroup differences: P = 0.27, I² = 17.4%.

Cholestasis (Analysis 8.34): Data show no difference in cholestasis overall (typical RR 1.26, 95% CI 0.86 to 1.84; participants = 616; studies = 5) and no significant subgroup difference according to age at commencement of amino acid; testing for subgroup differences: P = 0.22, I² = 34.5%.

Primary outcomes

Mortality to discharge (Analysis 9.1): Data show no difference in mortality (typical RR 0.90, 95% CI 0.69 to 1.17; participants = 1407; studies = 14) and no significant subgroup difference according to age at commencement of lipid; testing for subgroup differences: P = 0.79, I² = 0%.

Neurodevelopmental disability (Analysis 9.2): Studies that provided early lipid intake reported no difference in neurodevelopmental disability (typical RR 1.04, 95% CI 0.48 to 2.23; participants = 201; studies = 2).

Postnatal growth failure at discharge (Analysis 9.3): All studies reporting postnatal growth failure commenced early lipid. Data show a significant reduction in postnatal growth failure at discharge (typical RR 0.74, 95% CI 0.56 to 0.97; participants = 203; studies = 3).

Secondary outcomes

Days to regain birth weight (Analysis 9.4): Data show a reduction in days to regain birth weight overall (MD ‐1.14, 95% CI ‐1.73 to ‐0.56; participants = 950; studies = 13) and no significant subgroup difference according to age at commencement of lipid; testing for subgroup differences: P < 0.00001, I² = 92%. Studies commencing early lipid infusion reported a reduction in days to regain birth weight (MD ‐2.02, 95% CI ‐2.73 to ‐1.31; participants = 513; studies = 9). Studies with delayed lipid infusion ≥ 5 days reported no difference in days to regain birth weight (MD ‐0.57, 95% CI ‐2.04 to 0.91; participants = 199; studies = 2). Studies with no lipid reported an increase in days to regain birth weight (MD 2.04, 95% CI 0.58 to 3.50; participants = 238; studies = 2).

Maximal weight loss grams (Analysis 9.5): All studies reporting maximal weight loss in grams commenced early lipid infusion. Data show a decrease in maximal weight loss (MD ‐22.71, 95% CI ‐33.68 to ‐11.74; participants = 235; studies = 3).

Maximal weight loss per cent (Analysis 9.6): Data show no difference in maximal weight loss per cent overall (MD ‐0.33, 95% CI ‐1.61 to 0.96; participants = 288; studies = 4) and no significant subgroup difference according to age at commencement of lipid; testing for subgroup differences: P = 0.20, I² = 37%.

Weight gain to one month (Analysis 9.7): Data show a reduction in weight gain to one month overall (MD ‐1.50, 95% CI ‐2.56 to ‐0.44; participants = 373; studies = 4) with a significant subgroup difference according to age at commencement of lipid; testing for subgroup differences: P < 0.00001, I² = 94.9%. Studies commencing early lipid infusion reported no difference in weight gain to one month (MD 0.42, 95% CI ‐0.94 to 1.78; participants = 250; studies = 3). A single study with no lipid reported a reduction in weight gain to one month (MD ‐4.48, 95% CI ‐6.17 to ‐2.79; participants = 123) (Balasubramanian 2013).

Weight gain to discharge (Analysis 9.8): Data show no difference in weight gain to discharge overall (MD 0.76, 95% CI ‐0.02 to 1.54; participants = 291; studies = 4) and no significant subgroup difference according to age at commencement of lipid; testing for subgroup differences: P = 0.55, I² = 0%.

Linear growth to one month (Analysis 9.9): Data show a reduction in linear growth to one month overall (MD ‐0.16, 95% CI ‐0.26 to ‐0.06; participants = 245; studies = 2) with a significant subgroup difference according to age at commencement of lipid; testing for subgroup differences: P = 0.007, I² = 86.2%. A single study commencing early lipid infusion reported no difference in linear growth to one month (MD 0.00, 95% CI ‐0.15 to 0.15; participants = 122) (Clark 2007). A single study of no lipid reported a reduction in linear growth to one month (MD ‐0.27, 95% CI ‐0.40 to ‐0.14; participants = 123) (Balasubramanian 2013).

Head circumference growth to one month (Analysis 9.10): Data show no difference in head circumference growth to one month overall (MD 0.01, 95% CI ‐0.04 to 0.06; participants = 476; studies = 4) with a significant subgroup difference according to age at commencement of lipid; testing for subgroup differences: P < 0.00001, I² = 97%. Studies commencing early lipid infusion reported an increase in head circumference growth to one month (MD 0.07, 95% CI 0.02 to 0.13; participants = 353; studies = 3). A single study of no lipid reported a reduction in head circumference growth to one month (MD ‐0.38, 95% CI ‐0.51 to ‐0.24; participants = 123) (Balasubramanian 2013).

Head circumference growth to discharge (Analysis 9.11): All studies reporting head circumference growth to discharge commenced early lipid infusion. Studies that commenced early lipid infusion reported an increase in head circumference growth to discharge (MD 0.09, 95% CI 0.06 to 0.13; participants = 315; studies = 4).

Only studies that commenced early lipid infusion reported change in z scores.

Weight change z‐score to one month (Analysis 9.12): A single study that commenced early lipid infusion reported no difference in weight change z‐score to one month (MD ‐0.20, 95% CI ‐0.62 to 0.22; participants = 96) (Vlaardingerbroek 2013).

Weight change z‐score to discharge (Analysis 9.13): Studies that commenced early lipid infusion reported no difference in weight change z‐score to discharge (MD 0.01, 95% CI ‐0.33 to 0.36; participants = 207; studies = 2).

Weight change z‐score post discharge (Analysis 9.14): Studies that commenced early lipid infusion reported no difference in weight change z‐score post discharge (MD 0.13, 95% CI ‐0.26 to 0.52; participants = 201; studies = 2).

Head circumference change z‐score to one month (Analysis 9.15): Studies that commenced early lipid infusion reported an increase in head circumference change z‐score to one month (MD 0.27, 95% CI 0.08 to 0.46; participants = 231; studies = 2).

Head circumference change z‐score to discharge (Analysis 9.16): Studies that commenced early lipid infusion reported no difference in head circumference change z‐score to discharge (MD 0.18, 95% CI ‐0.15 to 0.50; participants = 207; studies = 2).

Head circumference change z‐score post discharge (Analysis 9.17): Studies that commenced early lipid infusion reported no difference in head circumference change z‐score post discharge (MD 0.25, 95% CI ‐0.14 to 0.64; participants = 201; studies = 2).

Days to full enteral feeds (Analysis 9.18): Studies reported no differences overall (MD ‐0.19, 95% CI ‐1.07 to 0.70; participants = 778; studies = 11) and no significant subgroup difference in days to full enteral feeds; testing for subgroup differences: P = 0.44; I² = 0%.

Late‐onset sepsis (Analysis 9.19): Data show no differences overall (typical RR 0.96, 95% CI 0.79 to 1.18; participants = 1255; studies = 15), no differences for subgroups, and no significant subgroup difference in late‐onset sepsis; testing for subgroup differences: P = 0.90, I² = 0%.

Necrotising enterocolitis (Analysis 9.20): Data show no differences overall, no differences for subgroups, and no significant subgroup difference in necrotising enterocolitis (typical RR 1.00, 95% CI 0.68 to 1.47; participants = 1301; studies = 14); testing for subgroup differences: P = 0.72, I² = 0%.

Chronic lung disease (Analysis 9.21): Data show no differences in chronic lung disease (typical RR 1.04, 95% CI 0.89 to 1.23; participants = 819; studies = 10) and no significant subgroup difference according to age at commencement of lipid; testing for subgroup differences: P = 0.07, I² = 63.1%.

Patent ductus arteriosus (Analysis 9.22): Data show a reduction in patent ductus arteriosus (typical RR 0.78, 95% CI 0.61 to 0.99; participants = 480; studies = 6) and no significant subgroup difference according to age at commencement of lipid; testing for subgroup differences: P = 0.26, I² = 25.5%.

Intraventricular haemorrhage (Analysis 9.23): Data show no differences in intraventricular haemorrhage (typical RR 1.12, 95% CI 0.74 to 1.69; participants = 341; studies = 3) and no significant subgroup difference according to age at commencement of lipid; testing for subgroup differences: P = 0.81, I² = 0%.

Severe intraventricular haemorrhage (Analysis 9.24): Data show no differences in severe intraventricular haemorrhage (typical RR 1.16, 95% CI 0.74 to 1.82; participants = 904; studies = 11) and no significant subgroup difference according to age at commencement of lipid; testing for subgroup differences: P = 0.54, I² = 0%.

Periventricular leukomalacia (Analysis 9.25): Data show no difference in periventricular leukomalacia (typical RR 0.55, 95% CI 0.24 to 1.25; participants = 720; studies = 7) and no significant subgroup difference according to age at commencement of lipid; testing for subgroup differences: P = 0.48, I² = 0%.

Retinopathy of prematurity (Analysis 9.26): Data show a reduction in retinopathy of prematurity (typical RR 0.44, 95% CI 0.21 to 0.93; participants = 269; studies = 4) and no significant subgroup difference according to age at commencement of lipid; testing for subgroup differences: P = 0.11, I² = 60.6%. Subgroup analysis showed a reduction for studies commencing early lipid infusion (typical RR 0.32, 95% CI 0.13 to 0.77; participants = 146; studies = 3). A single study of infants who did not receive lipid infusion reported no difference (RR 1.58, 95% CI 0.27 to 9.10; participants = 123) (Balasubramanian 2013).

Severe retinopathy of prematurity (Analysis 9.27): Data show no difference in severe retinopathy of prematurity (typical RR 0.96, 95% CI 0.56 to 1.63; participants = 672; studies = 8) and no significant subgroup difference according to age at commencement of lipid; testing for subgroup differences: P = 0.58, I² = 0%.

Cerebral palsy (Analysis 9.28): Studies that commenced early lipid infusion reported no difference in cerebral palsy (typical RR 4.00, 95% CI 0.89 to 17.97; participants = 122; studies = 2).

Developmental delay at ≥ 18 months (Analysis 9.29): Data show no difference overall in developmental delay at ≥ 18 months (typical RR 1.35, 95% CI 0.52 to 3.53; participants = 301; studies = 3) and no significant subgroup difference according to age at commencement of lipid; testing for subgroup differences: P = 0.37, I² = 0%.

Blindness (Analysis 9.30): Studies that commenced early lipid infusion reported no difference in blindness (typical RR 2.00, 95% CI 0.20 to 19.91; participants = 122; studies = 2).

Deafness (Analysis 9.31): A single study that commenced early lipid infusion reported no infant with deafness in either group (Vlaardingerbroek 2013).

Abnormal serum ammonia (Analysis 9.32): A single study that commenced early lipid infusion reported an abnormal serum ammonia > 122 μmol/L in one infant on higher amino acid intake (Blanco 2008).

Abnormal blood urea nitrogen (Analysis 9.33): Data show an increase in abnormal blood urea nitrogen overall (typical RR 2.77, 95% CI 2.13 to 3.61; participants = 688; studies = 7) with no significant subgroup difference according to age at commencement of lipid; testing for subgroup differences: P = 0.68, I² = 0%. Data show an increase in abnormal blood urea nitrogen in studies commencing early lipid infusion (typical RR 2.66, 95% CI 1.95 to 3.64; participants = 489; studies = 5). Studies that commenced lipid at ≥ 5 days reported a significant increase in abnormal blood urea nitrogen (typical RR 3.01, 95% CI 1.83 to 4.95; participants = 199; studies = 2).

Hyperglycaemia (Analysis 9.34): Data show a reduction in hyperglycaemia overall (typical RR 0.69, 95% CI 0.49 to 0.96; participants = 505; studies = 5) and no significant subgroup difference according to age at commencement of lipid; testing for subgroup differences: P = 0.08, I² = 68.1%. Studies that commenced early lipid infusion reported no difference in hyperglycaemia (typical RR 0.84, 95% CI 0.57 to 1.22; participants = 306; studies = 3). Studies that delayed lipid infusion ≥ 5 days reported a reduction in hyperglycaemia (typical RR 0.39, 95% CI 0.18 to 0.83; participants = 199; studies = 2).

Hyperglycaemia treated with insulin (Analysis 9.35): Data show no difference in hyperglycaemia treated with insulin (typical RR 1.24, 95% CI 0.93 to 1.66; participants = 534; studies = 5) and no significant subgroup difference according to age at commencement of lipid; testing for subgroup differences: P = 0.25, I² = 24%. Studies that commenced early lipid infusion reported no difference in hyperglycaemia treated with insulin (typical RR 1.29, 95% CI 0.96 to 1.73; participants = 420; studies = 4). A single study that delayed lipid infusion ≥ 5 days reported no difference in hyperglycaemia treated with insulin (RR 0.35, 95% CI 0.04 to 3.22; participants = 114) (Burattini 2013).

Hypoglycaemia (Analysis 9.36): Data show no difference in hypoglycaemia overall (typical RR 1.17, 95% CI 0.84 to 1.63; participants = 376; studies = 3) and no significant subgroup difference according to age at commencement of lipid; testing for subgroup differences: P = 0.45, I² = 0%.

Metabolic acidosis (Analysis 9.37): Data show no difference in metabolic acidosis (typical RR 2.05, 95% CI 0.94 to 4.47; participants = 305; studies = 4) and no significant subgroup difference according to age at commencement of lipid; testing for subgroup differences: P = 0.27, I² = 17.4%.

Cholestasis (Analysis 9.38): Data show no difference in cholestasis overall (typical RR 1.26, 95% CI 0.86 to 1.84; participants = 616; studies = 5) and no significant subgroup difference according to age at commencement of lipid; testing for subgroup differences: P = 0.11, I² = 63.7%. No study reported cholestasis in the absence of parenteral lipid administration.

Primary outcomes

Mortality to discharge (Analysis 10.1): High‐quality studies reported no difference in mortality (typical RR 0.74, 95% CI 0.42 to 1.29; participants = 568; studies = 5).

Neurodevelopmental disability (Analysis 10.2): Both studies reporting neurodevelopmental disability had methodological concerns.

Postnatal growth failure at discharge (Analysis 10.3): A single study at low risk of bias reported a reduction in postnatal growth failure at discharge (RR 0.59, 95% CI 0.39 to 0.88; participants = 50) (Can 2012).

Postnatal growth failure post discharge (Analysis 10.4): A single study reporting postnatal growth failure post discharge did not report the method of sequence generation and had imbalances between groups (te Braake 2005).

Secondary outcomes

Days to regain birth weight (Analysis 10.5): High‐quality studies reported no difference in days to regain birth weight (MD ‐0.49, 95% CI ‐1.87 to 0.89; participants = 94; studies = 2).

Maximal weight loss grams (Analysis 10.6): A single study at low risk of bias reported no difference in maximal weight loss in grams (MD 22.60, 95% CI ‐7.25 to 52.45; participants = 50) (Can 2012).

Maximal weight loss in per cent (Analysis 10.7): No high‐quality studies reported maximal weight loss in per cent.

Weight gain to one month (Analysis 10.8): A single study at low risk of bias reported no difference in weight gain to one month (MD 1.50, 95% CI ‐0.27 to 3.27; participants = 122) (Clark 2007).

Weight gain to discharge (Analysis 10.9): No high‐quality studies reported weight gain to discharge.

Linear growth to one month (Analysis 10.10): A single study at low risk of bias reported no difference in linear growth to one month (MD 0.00, 95% CI ‐0.15 to 0.15; participants = 122) (Clark 2007).

Head circumference growth to one month (Analysis 10.11): High‐quality studies reported an increase in head circumference growth at one month (MD 0.09, 95% CI 0.02 to 0.15; participants = 257; studies = 2).

Head circumference growth to discharge (Analysis 10.12): No study at low risk of bias reported head circumference growth to discharge.

Weight change in z‐score (Analysis 10.13; Analysis 10.14; Analysis 10.15): No high‐quality studies reported weight change in z‐score at any time point.

Head circumference change z‐score (Analysis 10.16): A single study reported an increase in head circumference change z‐score at one month (MD 0.37, 95% CI 0.15 to 0.59; participants = 135) (Morgan 2014). No study at low risk of bias reported head circumference change z‐score at any other time point (Analysis 10.17; Analysis 10.18).

Days to full enteral feeds (Analysis 10.19): High‐quality studies reported no difference in days to full enteral feeds (MD ‐0.20, 95% CI ‐1.60 to 1.20; participants = 169; studies = 3).

Late‐onset sepsis (Analysis 10.20): High‐quality studies reported no difference in late‐onset sepsis (typical RR 1.03, 95% CI 0.72 to 1.46; participants = 293; studies = 3).

Necrotising enterocolitis (Analysis 10.21): Studies at low risk of bias reported no difference in necrotising enterocolitis (typical RR 0.89, 95% CI 0.50 to 1.60; participants = 511; studies = 5).

Chronic lung disease (Analysis 10.22): Studies at low risk of bias reported no difference in chronic lung disease (typical RR 1.02, 95% CI 0.79 to 1.31; participants = 177; studies = 2).

Intraventricular haemorrhage (Analysis 10.23): A single study at low risk of bias reported no difference in intraventricular haemorrhage (RR 1.05, 95% CI 0.64 to 1.73; participants = 122) (Clark 2007).

Severe intraventricular haemorrhage (Analysis 10.24): Studies at low risk of bias reported no difference in severe intraventricular haemorrhage (typical RR 1.42, 95% CI 0.66 to 3.03; participants = 343; studies = 4).

Periventricular leukomalacia (Analysis 10.25): Studies at low risk of bias reported no difference in periventricular leukomalacia (typical RR 0.61, 95% CI 0.21 to 1.81; participants = 299; studies = 3).

Retinopathy of prematurity (Analysis 10.26): A single study at low risk of bias reported a reduction in retinopathy of prematurity (RR 0.16, 95% CI 0.04 to 0.67; participants = 75) (Can 2012).

Severe retinopathy of prematurity (Analysis 10.27): Studies at low risk of bias reported no difference in severe retinopathy of prematurity (typical RR 0.77, 95% CI 0.36 to 1.66; participants = 252; studies = 3).

Neurodevelopment (Analysis 10.28; Analysis 10.29; Analysis 10.30; Analysis 10.31): No study at low risk of bias reported neurodevelopmental outcomes (cerebral palsy, developmental delay, blindness or deafness).

Abnormal serum ammonia (Analysis 10.32): A single study at low risk of bias for biochemical outcomes reported an abnormal serum ammonia > 122 μmol/L in one infant on higher amino acid intake (Blanco 2008).

Abnormal blood urea nitrogen (Analysis 10.33): Studies at low risk of bias reported an increase in abnormal blood urea nitrogen (typical RR 12.29, 95% CI 1.66 to 90.79; participants = 183; studies = 2).

Hyperglycaemia (Analysis 10.34): No studies at low risk of bias reported hyperglycaemia.

Hyperglycaemia treated with insulin (Analysis 10.35): No studies at low risk of bias reported hyperglycaemia treated with insulin.

Hypoglycaemia (Analysis 10.36): No studies at low risk of bias reported hypoglycaemia.

Metabolic acidosis (Analysis 10.37): No studies at low risk of bias reported metabolic acidosis.

Cholestasis (Analysis 10.38): Studies at low risk of bias reported no difference in cholestasis (typical RR 1.21, 95% CI 0.67 to 2.17; participants = 249; studies = 2).