Blood pressure ranges via non-invasive and invasive monitoring techniques in premature neonates using high resolution physiologic data

Abbreviations

1 Introduction

More than 300,000 premature neonates (<37 weeks gestation) are born in the United States each year [1]. Premature neonates often require care in a neonatal intensive care unit (NICU) where routine measurement of blood pressure (BP) and mean arterial pressure (MAP) is performed. Blood pressure data currently published in the literature in the premature population is limited with regards to sample size, gestational age, and duration of analysis [2–6].

Clinical decision-making is often influenced by the presence of abnormal MAP, so knowing what is truly normal is crucial. Identifying and using accurate, evidence-based ranges can appropriately guide clinical care decisions and perhaps improve management strategies and avoid unnecessary treatments. We sought to establish BP reference ranges, specifically MAP ranges, in premature neonates based on gestational age (GA) and post-menstrual age (PMA) in a large cohort of neonates.

2 Patients and methods

The Institutional Review Board at the University of Virginia School Of Medicine approved this study. We performed a retrospective observational cohort study of all neonates admitted to the NICU at the University of Virginia Children’s Hospital between January 2009 and October 2015. We included neonates with GAs between 23 0/7 weeks and 34 6/7 weeks. There were no exclusion criteria. We collected BP data on a dedicated computer cluster and distinguished between non-invasive and invasive blood pressure measurements. Blood pressure values from neonates that did not survive to hospital discharge were included in the analysis. Non-invasive BP was measured intermittently using an automated BP cuff and analyzed in all neonates until a PMA of 38 6/7 weeks. The cuff measurement was recorded every two seconds based on a sample and hold technique by which the measured value was repeated until the next measurement. Upon admission the cuff is cycled every 15–60 minutes until neonatal stability, followed by cycling every 4 hours. In the subset of neonates with invasive monitoring in place (i.e. umbilical artery catheter or peripheral arterial line), BP was captured every two seconds from the bedside monitor by intra-arterial tracings. Analysis of invasive data was limited to the first week of age as most invasive lines are removed after seven days. To compare invasive versus non-invasive measuring techniques, we calculated Pearson’s correlation coefficient on a subset of simultaneous measurements.

Non-invasive values of MAP were stratified by both GA and PMA. Average values for each stratum were calculated to generate a heat map (Fig. 1a). We further sub-divided the data into four clinically-relevant GA groups: 23 0/7–25 6/7 weeks (group 1), 26 0/7–28 6/7 weeks (group 2), 29 0/7–31 6/7 weeks (group 3), and 32–34 6/7 weeks (group 4). Percentile curves and percentile reference tables were constructed for each GA group (Fig. 2a and Table 2). A heat map, percentile curves, and a percentile reference table were also constructed for neonates with invasive MAP data available during the first week of age (Figs. 1b, 2b, and Table 3).

OPEN IN VIEWER

Fig.1

Heat maps of average mean arterial pressure. 1a shows non-invasive mean arterial pressure based on gestational age and post-menstrual age and 1b shows invasive mean arterial pressure based on gestational age during the first week of age.

OPEN IN VIEWER

Fig.2

Mean arterial pressure percentiles for each gestational age group. 2a shows non-invasive mean arterial pressure percentiles as a function of post-menstrual age and 2b shows invasive mean arterial pressure percentiles displayed over the first week of age.

The administration of vasoactive medications and corticosteroids to neonates may modulate MAP values. Furthermore, administration of maternal antenatal steroids has been demonstrated to be associated with a higher MAP in neonates during the first 24 hours of age [7]. We employed stratified random sampling techniques to select a sub-sample of extremely preterm neonates (<28 0/7 weeks) to assess the incidence of utilization of these medications through review of the electronic medical record. We assumed a minimum rate of dopamine utilization among extremely preterm neonates to be 10% to select an appropriate sub-sample size [8, 9].

Continuous variables were compared using Student’s t-test or Wilcoxon rank sum testing as appropriate. Normality of continuous variables was assessed using Shapiro-Wilk testing. Categorical variables were compared using Chi square or Fisher’s Exact testing as appropriate. Type I error was set at 0.05 and a Bonferroni correlation employed for sub-sample analyses. All calculations were performed using R 3.3 (R Foundation for Statistical Computing) or Stata/IC 12.1 (Stata Corporation, College Station, TX).

We analyzed 2.05 billion non-invasive MAP data points from 1708 premature neonates. Within this patient cohort, 689 also had 129 million invasive MAP data points available during the first week of age. The median GA of the cohort was 31 weeks (interquartile range 28–33 weeks), and for the invasive group was 28 weeks (interquartile range 25–31 weeks). Table 1 shows patient demographics.

Blood pressure increased with both gestational and post-menstrual ages. Figure 1a shows a heat map of average non-invasive MAP values for each GA/PMA stratum. Squares in the lower left corner represent the most premature neonates at the beginning of their NICU course, and the darkest gray color signifies MAP of 35 mm Hg. Squares in the upper right corner represent less premature neonates toward the end of their stay, and the lightest gray color signifies MAP of 55 mm Hg. Figure 1b isolates the first seven days of neonates with arterial catheters.

Figure 2a shows line plots of MAP percentiles as a function of PMA for selected GA groups (Table 2 shows these data in a lookup table). The interquartile range, from 25th to 75th percentile, increases from 30 to 40 mm Hg in the most premature neonates just after admission to 47 to 60 mm Hg in least premature neonates near the end of their stay. Figure 2b isolates the first seven days of neonates with arterial catheters (Table 3 shows these data in a lookup table).

We analyzed 1465 simultaneously obtained invasive and non-invasive MAP measurements in 449 patients. There was moderate correlation (r = 0.57, 95% CI = 0.54, 0.61; p < 0.001), which is in line with previously published data [10]. Eighty-one percent of simultaneous invasive and non-invasive MAP measurements fell within 10 mm Hg of one another (Fig. 3).

OPEN IN VIEWER

Fig.3

Histogram distribution of differences between non-invasive (NBP) and invasive mean arterial pressure (IBP).

There were 372 neonates (22%) born at less than 28 0/7 weeks gestation included in our study. We performed a weighted stratified survey sampling by gestational age of 62 neonates (17%) randomly selected from this subset to assess the utilization of medications (e.g., dopamine, hydrocortisone) that may modulate blood pressure. The dopamine and hydrocortisone utilization for any period of time was 60% (±1.1%) and 26% (±0.9%), respectively. The average duration of dopamine infusion was 1.8 days (±0.12 days). Younger gestational age was associated with administration of dopamine (p = 0.02) though the clinical significance of this difference was minimal (24.9 weeks vs. 25.5 weeks gestational age). Gestational age was not associated with administration of hydrocortisone in this subset of extremely premature neonates (p = 0.13). There was no association between birth weight and dopamine or hydrocortisone utilization (p = 0.26 and p = 0.09, respectively).

Among these 62 neonates for whom electronic medical records were available (87%), antenatal steroids were administered to 81% of mothers with optimal timing of administration in only 42%, defined as the first dose given between 2 and 7 days prior to delivery [11]. The majority of the remaining mothers (42% of the total) received steroids less than 2 days prior to delivery.

We studied BP values in premature neonates throughout their NICU stay. We provide plots and tables of normative values for clinicians to use in the care of individual patients. Most notably, we have provided newfound knowledge of MAP at different gestational ages. There is a common belief that the MAP should roughly match the gestational age of the baby, however our findings suggest the MAP is greater than the gestational age of the baby. Based on an individual’s GA and PMA, the normative MAP range can be found easily, providing the clinician with adjunct knowledge in making intervention decisions, such as administering a fluid bolus or starting a vasopressor. Additionally, our data support the known phenomenon that BP increases substantially with both gestational and post-menstrual age. Prior, smaller studies reached qualitatively the same conclusions [2–6].

There are a number of iatrogenic factors that may modulate blood pressure in premature neonates. We found that more than half of extremely premature neonates received dopamine at some point during their NICU stay. However, among those who received dopamine, the average length of exposure was only 1.8 days. This is a short duration relative to the long hospitalizations typically required in neonates born at less than 28 0/7 weeks. Even so, there is the potential that we may have overestimated normal values in the extremely premature cohort. Due to the storage techniques of the data, it was not possible to separate out the blood pressure readings in patients receiving dopamine. Similarly, over half of neonates were exposed to at least one dose of antenatal steroids within the 7 days prior to delivery with the potential consequence of overestimation of normal values in this patient subset. It is also possible that confounders other than vasoactive use and antenatal steroid administration exist in this dataset.

There are some additional limitations as well. Our patient demographics, such as race, are not representative of all other institutions and it is possible some of these demographic factors unknowingly influence blood pressure values. The variability in subgroup size based on gestational age creates the possibility of sampling bias with a smaller number of extremely premature neonates in our sample relative to the number of premature neonates born at a more advanced gestational age. The distribution of gestational ages we observed is reflective of those that would be expected in most Level III NICUs thus minimizing any potential effect of bias [10, 12].

This study has substantial strengths. We have included over two billion data points in our analysis and our patient population includes GAs typically seen in neonatal intensive care units. We also included data until a PMA of 38 6/7 to encompass a more global picture of MAP values over the course of several weeks.

5 Conclusion

Analysis of blood pressure values from a large cohort of premature neonates provides useful heat maps, percentile curves, and reference tables in this vulnerable population. These results can be used at the bedside when assessing the critically-ill neonate.

Funding source

None

Financial disclosure statement

Randall Moorman, MD is the Chief Medical Officer of Advanced Medical Predictive Devices, Diagnostics, and Displays and owns stock. The remaining authors have no financial relationships relevant to this article to disclose.