The relationship between physician behaviors and blood gas values in the first hours of life--implications for "standards" of medical care for infants with respiratory distress.

It is standard practice for physicians to use blood gas (BG) evaluations when evaluating neonates with respiratory distress. In this study we addressed two questions: (1) What is the distribution of BG values in a population of infants receiving BG evaluation in the first 4 hours of life; and (2) How does the behavior of physicians correlate with BG values in these infants? We discuss the implications of our findings for claims about "standards" of medical care for newborn infants with respiratory distress. We reviewed medical records for 226 infants with birthweight > 2000 grams who were not intubated at the time of first BG determination. For 199 arterial samples, mean values were pH = 7.31 +/- 0.9 (SD); PaCO2 = 38.5 +/- 11.9 torr; PaO2 = 104 +/- 52 torr; and base excess (BE) = -6.5 +/- 3.8 mEq/L. These values did not differ significantly from previously published data for normal term infants without respiratory distress. However, the a/A ratio (0.45 +/- 0.19) for patients in our distressed population was significantly lower than reported for normal infants (0.65 +/- 0.10). For 186 infants admitted directly to our Newborn Intensive Care Unit, the elapsed time from birth to BG 1 was 1.07 +/- 0.64 hours. This value did not vary significantly as a function of severity of illness, assessed by pH, PaCO2, PaO2, a/A ratio, or BE. No blood gas parameter was simultaneously sensitive and specific for predicting subsequent mechanical ventilation. PaCo2 1 > 80 torr was associated with subsequent mechanical ventilation in 4 of 4 infants; however, the positive predictive value of PaCO2 1 was < 50% for levels below 80 torr, and only 4 of 22 infants eventually intubated were identified by a value of PaCO2 1 > 80 torr. The power of "abnormal" values of PaO2, a/A ratio, pH, or BE to predict subsequent intubation was even lower than PaCO2. Jurors in medical malpractice cases are instructed to define negligence as a deviation from the "skill and care ordinarily used in similar cases," and to determine the existence or absence of negligence guided by the testimony of "expert" witnesses. Recognizing that anecdotal recall of experience, even by "experts," may be inaccurate and is often systematically biased (the "Monday morning quarterback" phenomenon), we propose that the testimony of expert witnesses ought to conform, whenever possible, to a data-based description of medical care that actually is "ordinary used in similar circumstances". Our current observations suggest that (1) expert opinions of the "standard" to evaluate neonatal respiratory distress with a BG sample should reflect that the time scale is 1 to 2 hours, not 10 to 20 minutes; and (2) expert opinions that "abnormal" BG values either "require" or "preclude" intubation for most newborn infants with respiratory distress find little support in data. Clinical observation, not BG values, appears to be the most powerful "standard" by which physicians determine whether to initiate mechanical ventilation for newborn infants with respiratory distress.