Treatment of presumed hypotension in very low birthweight neonates: effects on regional cerebral oxygenation
- 1Department of Pediatrics, University of Arizona, Tucson, Arizona, USA
- 2Department of Pediatrics, University of Florida, Gainesville, Florida, USA
- Correspondence to David J Burchfield, Department of Pediatrics, University of Florida, College of Medicine, PO Box 100296, Gainesville, Florida 32610-0296, USA;
- Received 9 December 2012
- Accepted 18 April 2012
- Published Online First 10 July 2012
Context Previous studies have correlated poor neurological outcomes with hypotension. Treatment of hypotension in very low birthweight (VLBW) infants is common, and most often is based solely on the blood pressure measurement. Whether treatment improves cerebral oxygenation is unclear.
Objective To determine if treatment of hypotension in VLBW neonates results in an increase in cerebral oxygenation.
Patients and methods In this single centre observational study, neonates <30 weeks and <1500 grams, blood pressure and regional cerebral oximetry (rCSO2) with near infrared spectroscopy were continuously monitored and digitally recorded. If patients were treated for hypotension during the first 3 days of life, effects of treatment on blood pressure and regional cerebral saturation were determined.
Results Twenty-eight of 50 patients were treated by the medical team for hypotension, of which 22 had accurate data recorded for analysis. Both normal saline 10 ml/kg, and dopamine 2.5–5 mcg/kg per min significantly increased blood pressure, (saline 26.8±3.5 to 28.8±4.2 mm Hg, p<0.005; dopamine 27.6±1.9 to 29.5±3.2 mm Hg, p<0.02). Pre-treatment values of rCSO2 were similar to published normative values and treatment with either normal saline or dopamine had no effect on rCSO2.
Conclusion These results suggest that treating hypotension in VLBW neonates based solely on a blood pressure measurement of less than 30 mm Hg, while increasing blood pressure, may not increase cerebral oxygenation, possibly because many of these patients are in the autoregulatory zone for cerebral blood flow.
What is known on this subject
Up to 40% of very low birthweight neonates in the neonatal ICU are treated for hypotension, and hypotension is associated with poor developmental outcomes. Data conflict on whether treatment improves cerebral blood flow.
What this study adds
Treating VLBW neonates based solely on blood pressure less than 30 mm Hg does not increase cerebral oxygenation.
The circulatory system maintains oxygen delivery to the organs of the body. This inturn depends upon the oxygen content of the blood and the volume of blood delivered to the tissues. Oxygen content is easy to measure but measurement of blood flow is more difficult, so clinicians often use blood pressure as a surrogate.
The clinical relevance of hypotension in the very low birthweight (VLBW) preterm neonate is uncertain. Several studies document central nervous system morbidities associated with low blood pressure. In one, a mean blood pressure less than 30 mm Hg for over an hour in VLBW neonates was associated with severe haemorrhage, ischaemic cerebral lesions or death within 48 h and no severe lesions developed in patients with a MAP≥30 mm Hg.1 Similarly, Munro et al,2 using near infrared spectroscopy in extremely low birthweight infants, showed a breakpoint at approximately 30 mm Hg in the cerebral blood flow-mean arterial pressure autoregulation curve. This would suggest that under a mean BP of 30 mm Hg, cerebral blood flow is pressure passive, but above that measure, cerebral blood flow is maintained. Contrasting this notion, other studies suggest that blood pressure alone is unreliable in predicting cerebral perfusion.3 ,4 Therefore, the normal physiologic range for blood pressure to assure adequate cerebral blood flow and oxygen delivery is unknown in very preterm infants. Despite this, up to 40% of VLBW neonates are treated for hypotension in the first 3 days of life.5
Definitions and management of hypotension among VLBW neonates varies in neonatal intensive care units around the world.6 Variation in approaches to hypotension treatment may be due to several factors including the facts that blood pressure alone is a poor surrogate measure of end-organ perfusion, particularly cerebral perfusion,3 and hypotension7 and low cerebral blood flow8 are risks for poor neurodevelopmental outcomes.
The goal of treating hypotension should be restoration of organ perfusion, particularly cerebral perfusion, not just raising systemic arterial pressure. In that light, effects of various treatments differ. For instance, using 133Xe clearance to measure cerebral blood flow, Lundstrøm et al9 showed that although dopamine led to a higher blood pressure response compared with albumin infusion, albumin led to higher cerebral blood flow. Likewise, Osborn et al10 demonstrated that treatment of hypotensive neonates with dopamine led to an increase in systemic blood pressure without an increase in cerebral blood flow, as estimated by superior vena cava flow, but dobutamine improved cerebral blood flow by approximately 25% with no appreciable change in blood pressure.
Clearly, a different end point for therapy of hypotension needs to be investigated. Non-invasive methods for determining cerebral perfusion (indirectly and directly respectively), such as superior vena cava blood flow8 or middle cerebral artery doppler,2 are cumbersome and cannot be performed continuously. Near infrared spectroscopy (NIRS) uses the near infrared region of the electromagnetic spectrum and can be used to continuously estimate regional tissue oxygenation.11,–,13 Because tissue oxygen delivery is the ultimate goal of circulatory support, measurement of tissue oxygenation should be useful in assessment of cerebral autoregulation and obviate the need for measuring flow.
In this observational study, our aim was to determine if treatment for presumed hypotension improved both systemic blood pressure, an end point commonly used when treating hypotension, and cerebral oxygenation, arguably a more important end point.
Study criteria included all neonates admitted whose gestational age was less than 30 weeks, birth weight less than 1500 grams and arterial access in place. Study exclusion criteria included a cyanotic congenital heart lesion or suspected viability less than the 72 h study period. Informed consent was provided by a parent either prenatally or shortly after birth. The study was approved by the University of Florida's Institutional Review Board (IRB).
This was a prospective observational study. All diagnostic and treatment decisions were made by the medical team. Upon obtainment of informed consent, all patients were monitored (described below) for 72 h. If during that time patients were treated for hypotension, they were entered into the study. Data were analysed only on those patients who had clearly delineated times for treatment initiation and termination.
Monitoring blood pressure and cerebral oxygenation
Arterial saturations and blood pressure were continuously monitored (Agilent, Palo Alto, California, USA) as part of routine care. Additionally, regional cerebral oxygen saturation (rCSO2) measured by NIRS, using a neonatal cerebral probe centred across the central forehead, was measured using NIRS INVOS (Somanetics, Troy, Michigan, USA). At 30–60 s intervals, arterial blood pressure, percent arterial oxygen saturation and rCSO2 were recorded in a VitalSync computer (Somanetics, Troy, Michigan, USA). Data from the Agilent and INVOS began recording in VitalSyncTM within the first 6 h of life.
During the 72 h observation period, if a patient was treated for hypotension, the mean blood pressure and rCSO2 values for the 30 min immediately preceding treatment were averaged and compared with the 30 min epoch after completion of therapy (for volume administration) or a 30 min epoch 30 min after initiation of therapy (for vasopressors). The paired t-test was used to determine changes in cerebral oxygenation and systemic blood pressure after treatment. Values were considered significant if p≤0.05. PASW 18 (SPSS, Chicago, Illinois, USA) was used for statistical analysis.
Because only patients treated for hypotension were being enrolled in the study, and in our facility patients were typically treated for hypotension if the mean blood pressure were <30 mm Hg, the range of blood pressures of hypotensive patients would be narrow, and thus the SD would likely be small. Assuming a SD of 20% of the measure, and setting a 20% change in mean blood pressure as clinically important, 20 hypotensive patients would be required to show a treatment effect with α≤0.05, β≥0.8 and 2-tailed design. Previous studies have demonstrated that up to 40% of VLBW neonates are treated for hypotension,6 thus, we would need 50 monitored patients to accrue 20 hypotensive patients. We prospectively enrolled 50 patients to assure adequate power.
During the study period, 57 infants were admitted with a gestational age <30 weeks and birth weight <1500 grams. Seven were not enrolled. Of these, four patients were not enrolled due to mother's age being below that acceptable to the IRB to give informed consent, two refused to participate in the study and one patient was born with a birth anomaly that met exclusion criteria. Therefore, a total of 50 patients were monitored during the 72 h observation period.
Twenty-eight patients, each with a mean blood pressure less than 30 mm Hg, were treated for hypotension during the 72 h observation period, for which we could ascertain exact timing of treatment on 22. Of these 22, 20 received normal saline bolus infusion of 10 ml/kg (given over 15 to 60 min), 11 of these received dopamine after receiving normal saline and 2 received only dopamine. Clinical and demographic data comparing hypotensive versus non-hypotensive are shown in table 1.
Among those who were treated with normal saline, the mean post-natal age was highly variable at 312 min with a SD of 258 min. The average age of neonates treated with dopamine was 491 min with a SD of 337 min, also exhibiting great variability.
Enrolled patients had an average (±SD) gestational age 26.5 weeks (±1.8 weeks) and birth weight of 929 grams (±229 grams). Racial breakdown of study subjects included 46% African–American, 38% Caucasian and 16% other (including Hispanic, Asian and multiracial). Among the study subjects, 54% were female and 46% were male.
Blood pressure and rCSO2 response to treatment
Tables 2 and 3 and figures 1 and 2 show the responses of mean arterial blood pressure and regional cerebral oxygen saturation to treatment for hypotension. Figure 1 shows the response to treatment with volume expansion in the 20 patients, all receiving 10 ml/kg of normal saline as a bolus infusion. The mean blood pressure (±SD) for the group before treatment was 26.8±3.5 mm Hg, and at the completion of the volume infusion was 28.8±4.2 (p=0.004).
Also in figure 1, the pre-treatment and posttreatment mean blood pressure before and after dopamine infusion is seen in the 13 patients treated with dopamine, all of which were receiving 2.5–5 mcg/kg per min. In these patients, blood pressure increased from 27.6±1.9 to 29.5±3.2 mm Hg (p=0.02). We saw no difference in the blood pressure response after saline administration compared with after dopamine administration. The average percentage change in mean arterial pressure following volume expansion was 8.2%+/−12% and after dopamine was 7.2%+/−9.2% (p=0.6). However, rCSO2 did not change after treatment with either volume or dopamine (figure 2).
SaO2 and pCO2 response to treatment
We also evaluated the systemic oxygen saturations and partial pressure of carbon dioxide before and after treatment with normal saline (table 4) and dopamine (table 5). None of the comparisons were statistically different. Of note, the times at which pCO2s were obtained were highly variable as they were acquired as part of routine care.
Routine cranial ultrasounds were obtained within the first week of life and results were available for all neonates. While 14 had no intraventricular haemorrhages (IVHs), three had a grade I subependymal haemorrhages, four had grade II aIVHs and one had grade III aIVHs. Of the patients evaluated in this study, none died within the 72 h observation period. However, one died later from necrotising enterocolitis, one from pseudomonas sepsis, one from fungal sepsis and one from cor pulmonale.
Although neonatology has flourished as a paediatric specialty since the 1960's, the fundamental question of ‘what is an adequate blood pressure in VLBW neonates’ has not been answered. Several investigators have described the normal distribution of blood pressures over the first several days of life in low birthweight infants,14,–,18 and others have demonstrated untoward effects of low blood pressures, either mean blood pressure <30 mm Hg1 or less than the 10th percentile for birth weight and post-natal age.19 However, having a blood pressure within a statistical distribution of normal, especially in the sick, VLBW baby, may not be adequate to assure tissue oxygenation. Likewise, having a blood pressure below that norm does not guarantee inadequate tissue oxygenation either.
In this study, we have demonstrated that treatment for presumed hypotension, while leading to improvement in systemic arterial pressure, does not equate to an improvement in cerebral oxygenation. This may be due to the definition used for hypotension in VLBW neonates, which in our facility had been <30 mm Hg. In the 22 analysable patients treated for hypotension, the average rCSO2 pre-treatment was 71.8±7.8% and was 71.3±7.0% after treatment. This is very similar to the baseline rCSO2 values in non-hypotensive patients and similar to published normative values.20 This most likely proves that at the blood pressures being treated, the patients were still in an autoregulatory zone where cerebral perfusion was not compromised. These results substantiate those of Bonestroo et al21 who showed that treatment of hypotension, defined as blood pressure below the patient's gestational age, with volume expansion or dopamine did not improve regional cerebral oxygenation.
With the use of non-invasive near infrared spectroscopy, one can continuously monitor tissue oxygenation, particularly cerebral oxygenation, and use that information for treatment decisions. Care must be used to understand the pitfalls of this technology.22 Beat-by-beat signal to noise ratio is high which can be improved by a longer data acquisition period.23 More importantly, as shown by Kishi et al,24 interpatient variability, demonstrated by re-positioning the probe slightly on the forehead, showed measures differed by a mean of 2.7%, with 2 SD of 10.7%. NIRS primarily reflects cerebral venous oxygen saturation,25 and has good correlation with jugular venous oxygen saturation in children <10 kg.26 Because it reflects mostly venous saturation, measures will be dependent on tissue oxygen delivery, and tissue oxygen extraction. Therefore, interpretation of changes in the rCSO2 by NIRS must take into account any potential reasons for changes in tissue oxygen extraction, such as acute changes in metabolism brought on by seizures or anaesthesia.
Because we only monitored cerebral rCSO2, it may be that the patients in this study had compromised oxygen delivery to other organs such as intestine or muscle, and that redistribution of cardiac output spared cerebral tissue hypoxia. Hanson et al27 showed in dehydrated children that volume repletion increased regional oxygen saturation over the flank by 8% while cerebral rCSO2 remained unchanged. This would suggest that multi-site regional oxygen monitoring may have more promise in clinical decision making in regard to hypotension than a single cerebral site. However, in neonates such as those in our study, cerebral oximetry showed that cerebral oxygenation was not compromised in these patients, a presumption commonly made for the correlation between hypotension and poor neurological outcome.
In our study patients, 8 of 22 (36%) suffered from some degree of IVH. During that same year for patient accrual for this study, our centre reported an IVH rate of 15.7% to the Vermont Oxford Network. This doubling of IVH in our patient population demonstrates that these patients were most likely a sicker patient population than the aggregate reported to the network but does not answer whether the low blood pressure or the treatment of the blood pressure were causal in the haemorrhages.
These results suggest that treating hypotension in VLBW neonates based solely on a mean blood pressure less than 30 mm Hg, while increasing blood pressure, may not increase cerebral oxygenation, possibly because many of these patients are in the autoregulatory zone for cerebral blood flow. Instituting treatment for hypotension based solely on low blood pressure, without knowing if tissue oxygen delivery is compromised, places patients at risk for potential complications.28 ,29 In the future, improved non-invasive monitoring should be instituted for better clinical decision making when treating hypotension.
The authors acknowledge the contributions of Ms Cindy Miller, RN in her role as research nurse.
Funding Dr Garner was supported by a grant from the American Heart Association.
Competing interests Dr Burchfield obtained equipment used in the study from Somanetics, Inc, a company that has subsequently been purchased by Covidian, Inc. Neither Somanetics, Inc nor Covidian, Inc had a role in the design, data analysis or manuscript preparation of the submitted research.
Ethics approval University of Florida IRB.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement Minute-by-minute recordings of physiological variables for the first 3 days of life, in database format.