Historical Perspectives

Abstract

“The experimental hypothesis, in short, must always be based on prior observation.”—Claude Bernard (1)Modern adult intensive care started in Denmark in 1952 in response to the high mortality from the respiratory complications seen with a devastating epidemic of poliomyelitis. Anesthetists and physiologists created a new discipline when they collaborated to make sequential measurements of cardiorespiratory variables, then aggressively corrected abnormal conditions. (2) When modern neonatal intensive care emerged a decade later, it followed the same philosophy.The start of modern neonatology was stimulated in large part by an expanding understanding of: 1) the circulatory and respiratory physiology of adaptation to extrauterine life, 2) how those adjustments could fail, and 3) the pathophysiology of hyaline membrane disease. (3)(4)(5)(6)(7)(8)(9)(10) (The monograph by Dawes (3) gives an excellent overview of the relevant animal research). This research had reached a point where it could be applied to clinical care if the necessary physiologic measurements could be made in newborns. To this end, investigators began to catheterize the umbilical arteries of newborns to measure blood gas tensions and pH, with several investigators also measuring aortic pressures. (9)(11)(12)(13) Such invasive techniques were highly controversial initially, but they gradually became the accepted methodology for monitoring the sick neonate. (12)The use of oxygen in preterm infants who had cardiorespiratory distress was an important component of this change in approach to the sick infant. The standard had been to use no more than 40% oxygen to prevent retrolental fibroplasia (retinopathy of prematurity). However, it was obvious that this approach would be insufficient to relieve hypoxia in some infants and would produce hyperoxia in others. Avery and Oppenheimer (13) observed an increase in deaths due to hyaline membrane disease (respiratory distress syndrome) in preterm infants following the implementation of these standards. The new approach, which was gaining popularity in the 1960s, was to administer sufficient oxygen to maintain the Pao2 in the normal range. However, this could be achieved only by frequent measurements of Po2 in arterial blood, which required the placement of an umbilical arterial catheter. At this time, the technique for placing and maintaining peripheral arterial catheters had not been perfected, and reliable transcutaneous monitoring was in the distant future. Thus, placement of umbilical arterial catheters became standard practice in most of the few neonatal intensive care units (NICUs) that existed at the time. Prior to this, clinicians rarely used umbilical arterial catheterization. Although umbilical venous catheters had been in use for years for exchange transfusions in infants who had erythroblastosis fetalis, clinicians rarely took advantage of them for physiologic measurements.In 1964, we started placing umbilical arterial catheters in infants who had respiratory distress in the NICU at the University of California at San Francisco (UCSF). Later that year Bill Tooley (who started the NICU at UCSF and for whom the current NICU is named) and I asked ourselves, “Why not measure blood pressure in every infant with an indwelling umbilical arterial catheter?” It appeared possible with little, if any, added risk to the infant, so why ignore physiologic information that was used routinely in critically ill adults and older children? Attempts to measure blood pressure by indirect methods in neonates had had limited success and because many infants who had respiratory distress had been shown to have reduced peripheral perfusion (as noted below), the then available methods for indirect measurements were unlikely to be accurate or broadly applicable to the care of sick infants. (14)(15)In addition to the philosophical argument, there were three specific reasons to measure pressures through the umbilical arterial catheter. First, studies had shown that peripheral perfusion was subnormal in preterm infants with respiratory distress syndrome, (16) and Neligan and Smith (17) had found that blood pressure might be useful in monitoring the course of hyaline membrane disease. Second, such measurements could be used for the early detection of clot formation in the catheter. The type of catheter in use at this time was very prone to clot formation in the tip because it was made of polyvinyl chloride and only had a side hole near the tip. A clot often could be detected early by damping of the blood pressure wave form before other signs of clot formation (some of them disastrous) appeared. The third reason to measure pressures through the umbilical arterial catheter was to monitor for an adverse effect of acetylcholine, which sometimes was used at that time for treating severe hyaline membrane disease, but which occasionally produced a sudden marked drop in systemic blood pressure. (11)The system and routine for making these measurements was familiar to us because it was used regularly in the clinical and animal laboratories of the Cardiovascular Research Institute at UCSF. We employed a Grass polygraph that recorded on paper and Statham strain gauges (Fig. 1).Teresa Poirier, the highly creative head nurse of the NICU, was party to our conversation about measuring blood pressure in all infants with umbilical arterial catheters. Management of indwelling umbilical arterial catheters in neonates already had been made a nursing routine. Teresa insisted that if blood pressure monitoring was to be done well, it too must be in the hands of nursing, so she developed the nursing routine. It was added to the vital signs that the nurse recorded. The nurse recorded a 10- to 15-second strip of phasic pressure that provided systolic and diastolic pressures and heart rate, then reduced the frequency response to record mean pressure for another 10 to 15 seconds. The nurse wrote on the recorder paper the date and time, the infant’s name, and comments about the infant’s condition at the time (Fig 2), as well as entering the information in the bedside nursing record. Recordings were made hourly and could be made more frequently or even continuously if the infant’s condition was changing rapidly. We saved all of these recordings.We noticed that some patients had blood pressures that were much lower than we were used to seeing. Some of them appeared to be in shock due to acute blood loss that probably had occurred during the intrapartum period. Although following an individual baby’s changes in blood pressure was useful, it was obvious that we needed to define the range of normal blood pressure for the newborn to make the best use of the information. We also believed that “normal” probably would be lower in smaller, more preterm infants because studies in animals suggested that blood pressure increased with gestational age. (3) The problem was how to obtain data in truly normal infants, and the simple answer was that it could not be done. After struggling with this problem for some time, we realized that we did have data on two groups of patients who, although not normal, at least were healthy and that this information might provide a close approximation of normal.Where did these “normal” data come from? By 1965 we had begun placing umbilical arterial catheters soon after birth in preterm infants at high risk of developing respiratory distress. The need to measure arterial Po2 when providing high concentrations of oxygen to a preterm infant was noted previously. The earlier an attempt was made to catheterize the umbilical artery, the more likely it would succeed because the umbilical arteries constrict soon after birth. The desire to avoid the situation of needing to administer oxygen to an infant in whom we no longer could pass a catheter led to the practice of early catheter placement in high-risk preterm infants. A few of these infants never developed respiratory distress syndrome, and when the outcome was clear, the catheters were removed. Among other infants who had cardiorespiratory symptoms at birth and had been catheterized quickly, some improved rapidly, and the catheters were removed when it was clear that the infants would do well. In each of these situations, there was a period when blood pressure was recorded in infants who were “healthy,” and we decided to analyze these data to attempt to define “normal” blood pressure. Because of the circumstances of the data collection, we could examine the pressures only in the first 12 hours after birth. In a 30-month period, there were 45 “normal” infants among the 230 in whom umbilical arterial catheters were placed.Dr. Joseph Kitterman, at the time a research fellow in the Cardiovascular Research Institute, analyzed the data. He divided the infants into three weight groups and compared blood pressures measured during the first 12 postnatal hours. Systolic, diastolic, and mean pressures increased with increasing birthweight with each of the 12 time periods. However, there was little variation in the average pressures within each group over the 12 hours. That highlighted the need to use different normal values for newborns of different birthweights and indicated that “normal” for a particular weight was constant over the first 12 hours after birth. (18)Our experience with these measurements convinced us that mean pressure was the most useful measurement. Although the other pressures also were helpful, mean pressure best reflected perfusion pressure, and this single value made it easier to follow trends in an individual. Accordingly, Kitterman performed a more detailed analysis of mean pressure versus birthweight to create a graph of normal values for use by clinicians (Figure 4 in the original 1969 article). He found that a parabolic regression best fit the relationship. After publication, this figure was reproduced in several textbooks and came into common use.With time, one of the limitations of this research became a pressing problem. The “normal” data stopped at 1,000 g birthweight. At the start of this work, infants smaller than this weight usually died quickly and hardly fit even the loosest definition of “healthy.” However, over the next decade, survival increased at lower birthweights. Dr. Hans Versmold (then from the University of Munich and now at the Free University of Berlin) spent 1977 and 1978 as a visiting scientist at the Cardiovascular Research Institute and the Department of Pediatrics at UCSF and reviewed the experience with these smaller infants. Among 202 infants who had birthweights less than 1,000 g in whom umbilical artery catheters were placed, he identified 16 who were appropriate to add to the previous pool of “normal” data. He repeated the analysis of pressure versus birthweight for mean, systolic, diastolic, and pulse pressures. With this analysis, the best correlation was achieved with linear regressions for each of the pressures. (19) These were shown in Figure 4 of the 1981 paper, and that figure of “normal” blood pressure also has been widely reproduced in textbooks. A pharmaceutical company made it into a colored poster to be hung up in NICUs.[For the original article by Kitterman, Phibbs, and Tooley on aortic blood pressure in normal newborns, please see the data supplement.]Teresa Poirier, RN, John Clements, MD, and Joseph Kitterman, MD, gave me very helpful advice with this history.