Measuring gut perfusion and blood flow in neonates using ultrasound Doppler of the superior mesenteric artery: a narrative review

The gut is a relatively silent organ in utero but takes on a major role after birth for the absorption and digestion of feed for adequate nutrition and growth. The neonatal circulation undergoes a transition period after birth, and gut perfusion increases rapidly to satisfy the oxygen demand and consumption. If this process is compromised at any stage, preterm and fetal growth restricted infants are at particular risk of gut tissue injury secondary to hypoxia, leading to necrotizing enterocolitis. Feeding can also be a challenge in these high-risk groups due to gut dysmotility. Superior mesenteric artery (SMA) Doppler is a safe, bedside investigation that could rapidly aid clinicians with feeding strategies and in monitoring high-risk infants. This article aims to establish normal patterns of gut blood flow velocity in neonates using SMA Doppler and reviews how it might be used clinically in pathologic states.


Introduction
Following birth, there is a transition from fetal to newborn circulation, and thereafter, a newborn grows rapidly. Consistent forward flow through the superior mesenteric artery (SMA) to the gut is needed to perfuse the gut in order to support the main gut function, which is the absorption of nutrients for adequate growth. SMA Doppler ultrasound (US) allows us to assess the direction and velocity of gut blood flow; these factors affect a neonate's tolerance to feed. It may also be helpful in predicting and identifying the onset of gut-related diseases, such as necrotizing enterocolitis (NEC), and may guide clinicians on the best timing of blood transfusion (BT) for anemia, as this is currently under debate.

Basic principles of ultrasound
Doppler US has a high frequency and a short wavelength than the audible sound so it can be turned into a beam. A US probe emits a vertical beam and measures the returning echoes to formulate an image; if the echo is received from a moving tissue or cell (e.g., blood flow), then the transmitted and received frequencies are not equal. This is called the Doppler principle.
Blood flowing toward the probe has a higher frequency (red on color Doppler), and blood flowing away from the probe has a lower frequency (blue on color Doppler). Velocity, ultrasound frequency, and angle of insonation all influence the Doppler waveform. The higher the velocity, the higher the Doppler and US frequency; however, tissue penetration is better at lower US frequencies. A balance between the best achievable sensitivity to blood flow and tissue penetration is required. A pulsed wave Doppler (PWD) emits a single wave repeatedly fired at a particular frequency called pulse repetition frequency (PRF), and the echoes received from the moving blood cells will shift in frequency, which determines the blood velocity. PRF is also referred to as a scale on the echo machine. This is important where the blood velocity is not uniform [limited by changes in direction or at the level of valves/shunts such as patent ductus arteriosus (PDA)]. The highest detectable velocity is half the PRF (known as the Nyquist limit). PWD use is limited by aliasing where the blood velocity and beam/flow angle are greater than the Nyquist limit, which causes ambiguity with the results. For example, aliasing may occur if a low scale (low PRF) is used on a vessel with high velocity. This may cause the direction of blood flow to be registered incorrectly (known as color aliasing artifact). To prevent aliasing, the scale should be increased (which increases PRF) and the angle of insonation adjusted.

Ultrasound and Doppler use in medicine
US is widely used in many aspects of medicine and is useful for screening and prevention (e.g., mammograms), in diagnosis and assessing anatomy, and in assisting procedures (such as vessel catheterization). Doppler US produces flow velocity waveforms that allows physicians to monitor and identify changes in organ perfusion.
The peak systolic velocity (PSV), end diastolic velocity (EDV), and time-averaged mean velocity (TAMV) are used to calculate the pulsatility index (PI) and resistance index (RI) (1). See formula below: PI and RI are related measures of arterial pulsatility and can identify resistance in the blood vessel. The mean blood flow velocity (BFV) is often used to describe organ perfusion, and Doppler indices add information regarding pattern evolution over time.
Assessing the velocity of blood flow to a tissue alone does not directly describe the oxygenation of that tissue. Several other factors affect the blood's affinity for oxygen and its release to tissues (such as temperature, degree of acidosis, and proportion of fetal hemoglobin), which is well described by the oxyhemoglobin dissociation curve and Bohr effect.

Ultrasound and Doppler use in neonates
US is a safe and convenient bedside investigation and widely used in neonates in both clinical and research settings. Clinically, it is used regularly for the monitoring of the brain and is a valid bedside test to exclude intraventricular hemorrhage and parenchymal changes. US and Doppler US are also used to assess the neonatal skeleton, heart, abdomen, kidneys, and liver.

Gut blood flow and Dopplers
The primitive gut (made up of fore, mid, and hind gut) is supplied by three anterior aortic abdominal arteries. The coeliac trunk supplies the foregut (esophagus, stomach, and proximal duodenum as well as the hepatobiliary structures, pancreas, and spleen); the SMA supplies the midgut (the duodenum distal to ampulla of Vater, jejunum, and ileum; ascending colon and proximal two-thirds of the transverse colon); and the inferior mesenteric artery supplies the hindgut (distal third of the transverse colon, descending colon, and rectum) (2).
In utero nutrients pass via the placenta from the mother to the fetus, and the gut vasculature is relatively inactive. Achiron et al. investigated the changes in fetal SMA flow from 14 to 37 weeks gestation in normal fetuses; as expected, the intestinal resistance remained high throughout the pregnancy (as indicated by PI) (3).
Gut oxygen demand changes dramatically after birth. The relatively silent organ in utero requires a rapid switch to immediately absorb and digest nutrients for an animal to grow. Animal and human studies suggest that human gut maturation starts earlier in fetal life but is slower than other mammals postnatally (well beyond the age of weaning) (4,5). This could be impacted by preterm birth where the gut and lungs are immature. The fetus comes from a relatively hypoxic environment in utero, and a rapid and dramatic rise of arterial oxygenation is seen when breathing is established. If this does not occur (e.g., due to immature preterm lungs/birth asphyxia), then the gut does not function as expected [which explains the increased risk of NEC seen in preterm and fetal growth restriction (FGR) infants] (5). As the SMA irrigates the majority of the gut of interest in neonatology, few studies have interrogated the SMA blood flow velocities delineating the effects of various factors such as growth restriction and prematurity ( Table 1).

Gut disease in neonates
Gut disease in neonates is a particular challenge as we rely upon radiographic and laboratory findings to assess likelihood of diagnosis. However, many of these findings are not specific to that pathology, rather they appear in multiple gut pathologies. NEC is a potentially devastating inflammatory bowel condition, which affects mainly preterm infants (up to 10%) (6); however, the incidence differs depending on gestation, birth weight, and country at birth (7).
Gordon described that the diagnostic "pitfall" surrounding NEC is the inclusion of "NEC-like" illnesses in NEC data sets (8). For example, clinical signs of spontaneous intestinal perforation (SIP) are akin to the presentation of NEC. Often, diagnosis of SIP is only made surgically, so these infants may be incorrectly included as NEC if surgery is not warranted. Pneumatosis is pathognomonic for NEC, but is only present in 40% of the cases (9). Much work is being done to tighten NEC diagnosis, and clinically the focus lies on reducing the risk of NEC using feeding strategies. Currently, clinicians err on the side of caution if there are any abdominal concerns and stop feeding/ start antibiotics, which are not conducive to good weight gain or beneficial to the gut microbiome (10).

Effect of anemia and blood transfusion on gut blood flow
Up to 90% of extremely preterm infants require a BT to treat anemia in the neonatal unit. However, evidence for treating anemia in infants is conflicting; both anemia and BT may play a part in the development of gut tissue injury and NEC, possibly due to a period of gut hypoxia followed by a reperfusion injury. In addition, multiple studies suggest a causal relationship between BT and NEC [known as "transfusion-related NEC" (TRNEC)]. This relationship has been identified as a priority area for international researchers, and a recent meta-analysis of liberal and restrictive BT approaches showed no difference in the incidence of gut tissue injury (11).
SMA Doppler US may assist in identifying those infants who have reduced gut blood flow due to anemia. The basal intestinal vascular resistance drops rapidly after birth, and gut blood flow increases significantly during the first week of life (12). Several factors may impair this transition including anemia (13), need for cardiorespiratory support, and use of parenteral nutrition (12). Our group performed SMA Dopplers pre-and posttransfusion in 59 preterm infants and found higher pretransfusion SMA PSV in babies ≥8 days of age compared with those who were transfused in the first week of life; however, there were no changes in SMA PSV or EDV following BT (14). Several studies have demonstrated an attenuated response to feed immediately after BT, and this may last for 24-48 h (15). Currently, it is also not clear whether to withhold feeds or not during BT and the international WHEAT trial (WithHolding Enteral feeds Around packed red cell Transfusion to prevent necrotizing enterocolitis in preterm neonates) is likely to answer to this uncertainty (16).

Assessing intestinal blood flow
Performing a duplex scan of the mesenteric vessels was first suggested by Nicholls in the mid-1980s to aid the diagnosis of chronic intestinal ischemia in adults (17). Since the 1990s, the patterns of flow in SMA Doppler were established and deemed as highly accurate in detecting SMA stenosis/occlusion with a diagnostic accuracy of up to 90% (18)(19)(20)(21). By catheterizing the SMA and using dye dilution in adults, Norryd et al. showed a redistribution in blood flow after a meal and a corresponding drop in vascular resistance (22). Celiac artery flow did not demonstrate change in response to feed and did not affect SMA flow in any studies (17, 22, 23).

SMA Doppler in neonates: validity and what is already known
Cardiac output is distributed approximately equally to the upper and lower parts of the body in a healthy newborn (14), with abdominal organs accounting for the largest proportion of blood flow to the lower part of the body. After birth, there is a dramatic increase in intestinal growth and oxygen demand, and the SMA BFV increases significantly in the first few weeks of life to support this, particularly in preterm infants where the changes are more rapid. Where this rapid rise is not seen, infants are more likely to have intestinal dysmotility and feed intolerance (24,25).
Various studies have measured the SMA blood flow in clinically stable neonates ( Table 2). Stritzke et al. studied 21 healthy neonates born >36 weeks gestation to characterize brain, gut, and kidney BFV over time during the first 24 h. SMA PI and RI was elevated at birth and dropped significantly over time; it had the highest blood flow and Vmax and the largest variation compared with the other major arteries. The reported Vmax of 70.6 cm/s in SMA at 24 h corresponds with other reported normal values between 38 cm/s and 90 cm/s, and the wide variation is also similar to previous findings (probably related to feeds) (26, 27). These findings support the concept of an independent autoregulation of blood flow in different organs,

Factors that increase PSV
Factors that decrease PSV  Volume of feed did not affect post-feed response.
Daily fasted/pre-prandial SMA PSV and TAMV from Day 1-5 of life to identify clinical variables that relate to postnatal increase in gut BFV.
Prospective single-center study of 25 infants.
Abdominal NIRS (A-rSO 2 ) performed continuously for 72 h from enrollment.
Small sample size.
Only provided 3 days of measurements in the first 2 weeks of life.
No correlation between SMA Doppler and A-rSO2 changes 10 min post-feed.
SMA Doppler performed 3 times per day on 3 consecutive days (each set of 3 occurred immediately pre-, 10 min post-, and 60-120 min post-feed). Exclusion: congenital malformations, chromosomal anomalies, FGR, GI anomalies, previous NEC, need for inotropic support within 24 h, renal anomalies, or feed intolerance within 12 h of enrollment.
All infants were bolus fed.
Several factors impact intestinal blood flow, both intrinsic (including the cardiovascular status of the baby, mural and humoral control of vasculature) and extrinsic (volume and type of feed and certain drugs). It is important to optimize the factors we can control and understand better how they affect the rate of rise in SMA BFV especially in preterm infants (24). Research suggests that gestational age (GA), weight, FGR, feed, presence of PDA, and cardiovascular abnormalities all influence SMA BFV (30); and neonatal Dopplers correlate with disease severity and may differentiate those with anemia (27,31,32). The effects of these clinical factors will be individually discussed in more detail.

Influence of gut ischemia, gut tissue injury, and NEC on SMA Doppler
Compromising blood flow to the intestines causes hypoxia and ischemia. The cause of NEC is multifactorial, but perfusion and oxygenation of the gut certainly plays a part. Researchers have studied the effect of gut pathologies on the SMA blood flow ( Table 3). Murdoch et al. postulated that SMA Doppler performed on day 1 may identify infants at risk of NEC and dysmotility. They found that preterm infants with abnormal antenatal Dopplers who later developed NEC were more likely to have a raised PI on day 1 of life, suggesting that abnormal splanchnic circulation in fetal life or vasoconstriction in neonatal life increases the risk of NEC and this could be identified using SMA Doppler (33). However, this was not supported by a more recent study who found SMA Doppler on day 1 of life in preterm FGR infants to be a poor predictor of NEC (34).

Limitations of SMA Doppler
The normal pattern of gut blood flow is more difficult to establish than of the other organs as it is a motile organ and variation in intestinal wall tension affects vascular resistance inversely. The GI tract produces hormones that dilate intestinal vessels in response to enteric feed, which also contributes to increased blood flow (14).
Doppler assessment requires trained personnel and is prone to operator-dependent bias. The internal diameter of the vessel affects results and needs to be accurately measured (26). Neonatal blood vessels are <3 mm in diameter, bordering on the limit of resolution due to their size; fortunately, this is now less of a challenge with the availability of the more sophisticated color Doppler equipment (26). The angle of insonation affects results; Coombs et al. described an error of 10% in velocity for a 5% angle error when studying SMA flow, while Bel et al. found that an insonation angle of >15°caused underestimation in velocity (30).
Doppler provides a snapshot in time and does not allow continuous evaluation and does not provide a direct estimate of regional tissue perfusion (35). Other studies have reported difficulties with intestinal gas obscuring the view, infant restlessness, and variations in the course of proximal SMA that led to an unfavorable insonation angle (30).

Use of coeliac trunk Doppler in neonates
NEC is least likely to occur in the portion of gut supplied by the coeliac artery. There is no proven correlation between the coeliac artery and SMA Doppler results in infants with NEC (17, 25,26,33). Some studies suggest a "qualitatively similar but quantitatively smaller" response to feed compared with SMA Doppler (26) but a significant reduction in the flow volume of the celiac artery has been demonstrated following blood transfusion (36). However, coeliac trunk Doppler to assess gut blood flow in relation to gut tissue injury is unlikely to be clinically useful in neonates.

Effects of GA and postnatal age on SMA Doppler
Bel et al. studied normal values for SMA BFV in 128 preterm and term infants and identified a positive correlation between rising BFV (all parameters) with GA and body weight (30); this was similar to Murdoch et al. (33). Resistance (as evidenced by PI) was higher in lower gestations suggesting that degree of prematurity is associated with higher mesenteric flow resistance (33).
Our group studied infants in three postnatal age groups in the context of blood transfusion and found a significant positive correlation between pre-transfusion PSV and postnatal age; this difference was particularly apparent when comparing infants in the first week of life vs. >28 days. This was attributed to the circulatory adaption after birth and possibly the presence of PDA. However, higher PSV with postnatal age remained significant when corrected for GA, birthweight (BW), and the presence of PDA (37).

Effects of FGR and SGA on SMA Doppler
Fetal brain sparing or cardiovascular adaptation that occurs during FGR is to the detriment of the "non-essential" fetal organs such as the gut and is particularly evident where abnormal umbilical Doppler results are present (38,39). Infants predisposed to gut hypoxia develops feed intolerance and therefore are at increased risk of developing NEC (38,(40)(41)(42)(43). These infants are also more likely to require BT for anemia (44).  Exclusion: multiple pregnancy, major congenital and chromosomal anomalies, CHD, shock, on inotropes, previous NEC.  Several studies have demonstrated a positive correlation between gut BFV and body weight (30,33,45). FGR infants appear to have different waveforms compared with their appropriately grown for gestational age (AGA) counterparts (30). Infants with FGR have high resistance patterns of flow in the first few days of life particularly if they later develop NEC (34), suggesting that in utero resistance/vasoconstriction persists after birth (33).

Effects of anemia and BT on SMA Doppler
In many forms of anemia, the hematocrit is low causing reduced viscosity and increased velocity, and our group have demonstrated higher SMA PSV in anemic preterm infants during the first week of life (46). Anemia may not impair post-prandial increase in mesenteric blood flow but causes a greater than normal increase in velocity in response to feed (47). Blood viscosity increases following BT (up to 20%), and as expected, several studies have demonstrated a concurrent significant reduction in the mean PSV (15, 46,47). However, it is not clear how long these changes persist for after BT (15). The effects of anemia and blood transfusion on SMA Doppler are summarized in Table 3.

Effects of sepsis on SMA Doppler
In the septic state, acute hypotensive shock may occur, and vascular bed arterioles constrict in response to pathologic decrease in systemic vascular resistance. Redistribution occurs that reduces splanchnic flow, predisposing infants to gut tissue hypoxia and injury as early as day 1 of life (48). Animal and human models suggest a rise in RI, and a drop in EDV is seen in culture-positive sepsis (49).
The risk of NEC increases with raised PI and falls with high EDV, and infants with risk factors for NEC tend to have higher resistance patterns as early as day 1 of life (48). Studies using SMA Doppler during NEC are outlined in Table 3. However, in active NEC/sepsis, this resistance reduces due to dilatation of the vascular bed and inflammation (26, 33).

Effects of feeding on SMA Doppler
Feeding is often difficult in preterm and FGR infants, and clinicians approach cautiously (43). It is the introduction of colostrum after birth that increases gut blood supply and encourages development of vascular bed. Human gut maturation starts early in fetal life, but is slower than other mammals and does not fully mature until beyond the time of weaning onto solid foods (5). Both animal and human models demonstrate a rapid rise in SMA BFV and a drop in vascular resistance on Doppler in the first week of life as feeding is established (5,45,50). This is clinically significant as a more rapid rise in SMA BFV reduces gut dysmotility, and a high resistance, reversed or turbulent flow pattern across SMA, is associated with a delay in establishing feeds (5,50).
In normal infants, enteral feed stimulates intestinal motility and release of circulating vasoactive substance (e.g., CCK P2, secretin, and gastrin), which increases intestinal blood flow. It is accepted that PSV is higher in fed babies, and PSV rises further following a feed. Preterm infants are able to regulate SMA BFV in response to milk feeds, and increases in velocity have been shown to be positively correlated with feeding tolerance (26, 27,52). Response may also be affected by the feed volume, frequency, and type (11,25,50). Havranek et al. demonstrated that the total volume of enteral feed (regardless of human or formula milk) was significantly related to a daily increase in TAMV (12). Lane et al. showed persistent hyperemic state in infants fed by hourly bolus compared with every 3 h bolus feeds with significantly higher pre-prandial PSV (70 vs. 53 cm/s). Those receiving breast milk had a longer but smaller change in amplitude compared with those receiving preterm formula (51), suggesting hourly bolus feeds cause persistent hyperemic state,  ↑ NEC in infants with ↓EDV and ↑PI.
Used both suspected and confirmed cases of NEC.
Inclusion: <24 h of life at study entry. As risk of NEC↑ -> TAMV↓.
Clinicians were blinded to Doppler results.
As GA ↑ -> ↑PI and ↓EDV. Exclusion: congenital anomalies or proven sepsis Infants were monitored for NEC to see if SMA Doppler on Day 1 can predict NEC.
SMA, superior mesenteric artery; PSV, peak systolic velocity; EDV, end diastolic velocity; EDF, end diastolic flow; A/REDF, absent/reversed end diastolic flow; PI, pulsatility index; RI, resistance index; TAMV, time-averaged mean velocity; BFV, blood flow velocity; FGR, fetal growth restriction; SGA, small for gestational age; AGA, appropriately grown for gestational age; GA, gestational age; CGA, corrected gestational age; PROM, prolonged rupture of membrane; NEC, necrotizing enterocolitis; PDA, patent ductus arteriosus; CHD, congenital heart disease; BT, blood transfusion. and the composition of feed and feed interval affects gut perfusion (26, 51). Coombs et al. made similar conclusions with regard to bolus feeds that may persist for up to 2 h (26). Hypoxia/ischemia of intestine may alter the interaction between intestinal motility and the release of vasoactive substances in response to feed since infants who showed poor vasomotor response were symptomatic of gut dysmotility, and almost 50% of these infants went on to develop sepsis (25). This is outlined in more detail in Table 4.

Effects of PDA on SMA Doppler
The presence of a PDA is inversely related to gestational age, and while exact numbers differ, PDA is present in at least 50% of infants born at <28 weeks (52) but may be as high as 90% in <24 week infants and 80% in 25-28 weeks (53). A significant proportion of these is hemodynamically significant PDA (hsPDA) and requires treatment for closure (53,54). An hsPDA is when a left to right shunt leading to pulmonary overcirculation and a left heart volume overload are seen. Clinically, infants might present with ventilator dependence due to pulmonary edema and ineffective gas exchange. An hsPDA and the resultant ductal steal lead to retrograde diastolic flow in the abdominal aorta, and a low antegrade or retrograde diastolic flow in systemic arteries leading to systemic hypoperfusion (55). A low PSV or reversed end diastolic flow (EDF) with raised PI has been reported in anterior cerebral artery and renal and mesenteric arteries, and the degree of resistance is proportional to the size of the shunt (56). The size of the shunt is proportional to the degree of morbidity (increased risk of intraventricular hemorrhage, NEC, acute kidney injury, and death).
The association between PDA and the risk of NEC is complex. Gut hypoxia has been demonstrated to increase the risk of NEC, and the presence of both PDA and the drugs that encourage PDA closure may cause gut hypoperfusion. Prophylactic surgical ligation may reduce the risk of NEC (57), but more recent RCTs revealed no difference in NEC rates in infants receiving prophylactic pre-symptomatic medical treatment (58). As such, prophylactic treatment of PDA is no longer used, and clinicians approach treatment more cautiously (59,60).
McCurnin and Clyman found that the presence of a PDA limits the usual increase in post-prandial SMA flow in preterm baboons (61), and similar effects were seen in neonatal studies (55,62,63) (Table 4). Our group showed significantly higher PSV in preterm infants with a PDA if they were receiving at least half of the total fluids enterally compared with those who received <50%, but EDV was similar in both feed groups (37), indicating that interrupting the SMA blood flow in response to feed may impede normal digestion (62).

Effects of congenital heart disease on SMA Doppler
Infants with left ventricular outflow obstruction of any cause (e.g., critical aortic stenosis, coarctation, or interruption of aorta) may be at risk of abnormal BFV of the SMA (reduced PSV and mean velocity and increased resistance) (28). While the ductus arteriosus is open, some blood flow is maintained to the lower extremities. However, when the duct closes, BFV in the distal vessel falls and resistance rises dramatically, causing profound hypoxia to all distal organs including the gut. This can cause a devastating, potentially fatal NEC due to a rapid and profound reduction in blood supply and oxygen delivery and causes gut ischemia.

Other biomarkers of gut perfusion and SMA Doppler
The use of NIRS in monitoring gut oxygenation in preterm infants is still being understood. While SMA Doppler reflects blood flow, NIRS measures oxygen delivery (and subsequent oxygen extraction). These two aspects of oxygen delivery to an organ are inherently different while still being related. Nevertheless, their association is an important one to understand. Gillam-Krakauer et al. found that the change in splanchnic regional oxygenation (SrSO 2 ) measured using NIRS correlated well with the changes in SMA BFV (using Doppler US) before and after feeds, suggesting that SrSO 2 reflects intestinal blood flow and can be used to assess intestinal perfusion (64). Our group performed concurrent SMA Doppler and gut NIRS in investigating the effects of anemia and the changes in response to BT and found that BT improved intestinal oxygenation but did not alter mesenteric BFV, suggesting these two measures may be more clinically useful when used together rather than independently (37).
Lactic acidosis is a non-specific marker of tissue ischemia and is easy to identify on bedside blood test. It would be useful for future work to include SMA Doppler measurements during lactic acidosis, as well as its relationship with known biological markers of gut injury such as intestinal fatty acid binding protein (iFABP), the latter of which is currently being investigated by our group in both FGR and AGA preterm infants.

Conclusion
SMA Doppler is a safe and reliable bedside test to assess gut blood flow for unwell and premature neonates. Doppler can assist in identifying adequate transition from fetal to newborn circulation and may be useful in high-risk infants (e.g., FGR). High resistance patterns of flow persist after birth for approximately 24 h in these infants; perhaps SMA Doppler could identify when this resistance is dropping and guide clinicians when to safely start feeding.
The risk of NEC increases with raised PI and falls with high EDV, and infants with high-risk factors for NEC have high resistance flow patterns on day 1 of life. While an early use of SMA Doppler may identify infants who are at high risk of developing NEC, there is a lack of evidence that the use of  Aimed to investigate GI side effects of indomethacin (e.g., NEC).
Hsu et al.
Prospective dual-center study of 25 infants.
Small sample size.
Did not investigate beyond first week of life. Inclusion: inborn infants <30 weeks.
Size of PDA α resistance and 1/α velocity. Abnormal organ blood flow = abnormal indices in ≥2 organs on same day of life or ≥2 abnormal blood flow on different days of life Exclusion: chromosomal or major congenital anomaly, complex congenital heart disease and perinatal asphyxia, severe IVH/SIP/NEC/ culture-positive sepsis in first week of life. Abnormal blood flow considered as reversed/absent PDA ≥2 mm -> 8-fold ↑ in RI.

Author contributions
CM and SB contributed to the study design, screened the search results, reviewed all included studies, drafted the initial manuscript, and reviewed and revised the manuscript. JB contributed to the study design and reviewed and revised the manuscript. NA conceptualized this systematic review and reviewed and revised the manuscript. All authors contributed to the article and approved the submitted version.