Objectives. To learn whether US obstetricians and pediatricians accurately estimate rates of survival and freedom from handicap in preterm infants and to learn whether their knowledge and attitudes influence their choice of interventions that may enhance survival.
Methods. A cross-sectional survey of obstetricians and pediatricians practicing in the United States was performed using a pretested questionnaire designed to identify their knowledge regarding survival and handicap-free rates of infants born at 23 to 36 weeks of gestation. At each week of gestation, they were asked whether they would provide specific therapeutic interventions to either the expectant mother or infant. Survival and handicap-free rates were compared with published national rates. Obstetricians and pediatricians were divided into an optimists group and a pessimists group, based on their estimates of survival. The rates at which each group used therapeutic interventions were compared.
Results. Both obstetricians and pediatricians underestimated survival rates from 24 through 35 weeks of gestation and freedom from serious handicap from 23 through 36 weeks of gestation. On the average, optimists accurately predicted neonatal survival. Obstetricians who underestimated neonatal survival would less often administer antenatal corticosteroids, perform a cesarean section for fetal distress, and transfer a mother to a tertiary center. Pediatricians who underestimated neonatal survival would less often use mechanical ventilation, cardiopulmonary resuscitation, inotropes, intravenous fluids, thermal support, and oxygen supplementation.
Conclusion. Physicians underestimate survival and freedom from handicap in preterm infants. Underestimation of outcome is associated with restriction in the use of appropriate interventions.
Decision-making by obstetricians and pediatricians regarding perinatal interventions relies heavily on their knowledge and attitudes. The reported misperceptions about neonatal survival and eventual quality of life for preterm infants may result in less than appropriate care for pregnant women and their infants.1,,2Underestimation of neonatal outcome is especially important because advances in the care of preterm infants generally have improved survival without increasing rates of serious handicap.3–8We have reported that both pediatricians and obstetricians in Alabama currently underestimate rates of survival and freedom from handicap in preterm infants.9,,10 Through analysis of hypothetical management by surveyed obstetricians and pediatricians in our state, it was demonstrated that physicians with inaccurate knowledge were less aggressive in attempting to salvage potentially viable fetuses and neonates.9,,10 This study was designed to determine if US physicians accurately estimate survival and freedom from handicap of preterm infants. The study also sought to determine whether knowledge and attitudes about survival rates of preterm infants affect physicians' treatment practices.
A questionnaire that asked pediatricians and obstetricians to estimate survival and the percentage of survivors who would be expected to be free of major handicap for preterm infants born under optimal conditions at a perinatal center at 23 through 36 weeks of gestation was developed and pretested. A hypothetical case of a healthy primigravid woman in preterm labor with intact membranes was given to the obstetricians. The obstetricians were asked which interventions they would undertake at each week of gestation, including the administration of corticosteroids, administration of tocolytics, monitoring of the fetus in labor, cesarean section for fetal distress, and transfer of the mother to a center capable of ventilating sick newborns (assuming their own hospital did not have that capability). The pediatricians were asked which interventions they would undertake at each gestational age including thermal support, oxygen by hood for cyanosis, intravenous fluids for hydration and/or nutrition, inotropic agents for treatment of hypotension, mechanical ventilation for respiratory distress, and cardiopulmonary resuscitation.
In 1993, expedited human subject approval was obtained from the University of Alabama at Birmingham Institutional Review Board. Cover letters, questionnaires, and self-addressed stamped envelopes were mailed to 1158 physicians board-certified in obstetrics–gynecology randomly selected from the American College of Obstetricians and Gynecologist Directory and to 699 physicians randomly selected from the American Academy of Pediatrics Fellowship Directory who were not practicing a pediatric subspecialty. These sample sizes were chosen based on response rates in the local pretesting surveys and were randomly selected based on geographic distribution for each specialty. Questionnaires were serially coded to allow for the second mailing to nonresponders. Study participants were told in the cover letter the reason for the serial number coding and assured anonymity of their responses. Two weeks after the first mailing, a second mailing was sent to all nonrespondents. Participants who indicated subspecialty practice or faculty status in subspecialties were eliminated from analysis.
Multicenter data obtained from the 1988–1989 National Institute of Child Health and Human Development (NICHD) Neonatal Network11 and from the 1982–1986 March of Dimes Multicenter Prevention Trial12 were used for survival rates comparisons. NICHD data were chosen for gestational ages 23 through 30 weeks, inclusive, because at the time of the questionnaire, they were the most recent multicenter survival data of extremely preterm infants published. However, because the NICHD Neonatal Network study reported outcomes of infants who were born at <1500 g birth weight, only data of infants up to 30 weeks of gestation were used, because infants of >30 weeks of gestation were more likely to have been small for gestational age. Survival rates for infants of 31 through 36 weeks of gestation were obtained from the earlier March of Dimes Trial that entered all infants born before 37 weeks of gestation regardless of birth weight.12
Handicap-free rates from a summary of outcome studies published between 1978 and 1984 were used.13 This reference was chosen because handicap rates were reported by gestational age rather than birth weight as is frequently done. Handicap-free was defined as not having a major handicap, including cerebral palsy, mental retardation (intelligence quotient: <70), developmental delay (developmental quotient: <70), blindness, deafness, or severe failure to thrive. More recent reports show even lower handicap rates, but the data could not be used in the current study because they were analyzed by birth weight and not by gestational age.
To determine whether the estimates of survival rates correlate with the willingness to use therapeutic interventions, participants were subdivided into optimists and pessimists based on their estimates of survival rates at each gestational age. An optimist group was formed at each gestational age from those physicians who estimated survival above the median response. Similarly, a pessimist group for each week of gestation was composed of those who estimated survival below the median response. The groups were compared regarding their willingness to intervene with therapeutic interventions.
The estimates of survival and freedom from handicap at each gestational age were compared with rates obtained from the literature review (actual) using unpaired t tests performed at each gestational age between 23 and 36 weeks of gestation, inclusive. To assess the relationship between survival estimates and willingness to use therapeutic interventions, differences between the optimist and pessimist groups' percent use of each intervention at each gestational age were tested for significance using χ2analysis.
Using relative risk ratio analysis, which is commonly applied to exposure/disease dyads, optimists and pessimists were compared as to their likelihood of using each therapy. Optimists and pessimists were considered exposures and the use or lack of use of each therapy was considered the disease. Using this construct, 2 × 2 tables were formed and a relative risk ratio was calculated and termed likelihood of use. In keeping with relative risk ratio analysis, 95% confidence interval (CIs) were also determined for each therapy at each gestational age.
Three hundred seventy-nine (33%) obstetricians and 362 pediatricians (52%) responded after the 2 mailings. Ninety-four percent of the obstetrician responding and 75% of the pediatricians responding were able to be included after removal of those indicating subspecialty practice. Responding obstetricians were on average 45 years of age and 76% male. The percentages of obstetricians associated with each nursery level were: 47% with level III, 41% with level II, and 12% with level I centers. Pediatricians who responded were also on average 45 years of age and predominately male (65%). Obstetricians underestimated survival at each week from 23 through 36 weeks of gestation (each P < .05; Fig 1). Pediatricians significantly underestimated survival rates at each week from 24 through 35 weeks of gestation (each P < .05; Fig 1). In addition, both groups of physicians significantly underestimated handicap-free rates of infants from 23 through 36 weeks of gestation (eachP < .05; Fig 2). At the more advanced gestational ages, estimates of both pediatricians and obstetricians were progressively closer to the actual rates. Optimists accurately estimated survival rates and handicap-free rates, whereas pessimists significantly underestimated both rates.
Among pediatricians, the optimist and pessimist groups differed in their stated willingness to use therapeutic interventions. The optimist group was 1.1 (95% CI: 1.01–1.17) to 1.8 times (95% CI: 1.22–2.67) more likely to use mechanical ventilation for infants between 23 and 27 weeks of gestation. Inotropic support was 1.1 (95% CI: 1.04–1.25) to 1.6 times (95% CI: 1.05–2.55) more likely to be used by optimists for infants 23 to 27 weeks of gestation (Fig 3). Thermal support and oxygen were more likely used by optimists at 24 weeks of gestation 1.9 (95% CI: 1.01–1.18) and 1.22 (95% CI: 1.07–1.39) times more likely, respectively. Cardiopulmonary resuscitation was 1.24 (95% CI: 1.02–1.3) to 1.35 (95% CI: 1.03–1.77) times more likely to be used by optimists at 24 and 25 weeks of gestation. Intravenous fluids were likely to be given 1.24 (95% CI: 1.11–1.39) to 1.50 (95% CI: 1.26–1.77) times more often by optimists at 24 and 25 weeks of gestation (Fig 3). On average, the optimists indicated a willingness to use the following interventions earlier than the pessimists; mechanical ventilation at 24 versus 25 weeks (P = .004), inotropes at 24 versus 25 weeks (P = .0057), cardiopulmonary resuscitation at 25 versus 26 weeks (P = .01), and intravenous fluids at 23 versus 24 weeks of gestation (P = .017).
Among obstetricians, optimists and pessimists differed in their willingness to perform a cesarean section for distress, administer steroids, and transfer of the mother in preterm labor. Optimists were 1.3 (95% CI: 1.1–1.6) to 2.8 (95% CI: 1.23–6.21) times more likely to perform a cesarean section for distress in infants between 23 and 25 weeks of gestation (Fig 4). Antenatal corticosteriod use was 1.3 (95% CI: 1.1–1.6) times more likely at 24 weeks and 1.2 (95% CI: 1.0–1.3) times more likely at 25 weeks of gestation. Transfer of a mother in preterm labor to a tertiary center was 1.3 (95% CI: 1.0–1.6) times more likely at 23 weeks of gestation (Fig 4). Optimists and pessimists did not significantly differ in their use of tocolytics or fetal monitoring during labor.
The purpose of this study was to determine whether physicians involved in the care of preterm infants have an accurate knowledge of survival and handicap rates and whether this knowledge affects their willingness to use therapeutic interventions. Overall, we found that both pediatricians and obstetricians consistently underestimate both survival and freedom from handicap of preterm infants. Additionally, those physicians who thought infants would have a poorer outcome were less willing to use some therapeutic interventions at the earliest gestational ages.
To reach many physicians from a broad geographic distribution, a questionnaire was used. This questionnaire was completed by 40% of its targeted audience. Although a higher response rate would have been preferred, this rate is comparable or better than that of similar studies using a survey format.14–16 However, the accuracy of a questionnaire to assess knowledge, attitudes, and practice, is limited by several factors including the ability to reach the appropriate target audience and the ability of the individual respondent to interpret the questions and answer appropriately.17 Survey-based data collection is also subject to other potential biases. Study participants may say what they believe is correct, although it may not agree with what they do in practice. A respondent's answer to 1 question may be dependent on his or her answer to another. However, the questionnaire used in this study is very similar to the 1 we have used extensively in the past and results of the current national study are similar to those of previously published single state data.1,,9,10
Obstetricians and pediatricians take care of a large number of mothers in preterm labor and their infants, respectively. For example, nearly one third of very low birth weight infants initially receive care outside of a tertiary care center.18 The obstetricians and pediatricians were surveyed regarding interventions that they may have to initiate and not interventions before any necessary transfer to a tertiary center.
Although nonrespondents were not able to be compared with respondents with regard to demographics, the results are consistent with previously published data from our statewide survey in which responders and nonresponders were found to be comparable with regard to gender makeup. In the statewide survey, responders were board-certified later and tended to work in areas without neonatologists. These differences would likely make nonrespondents more likely to underestimate outcome than respondents. However, it is possible that in the current study, the nonresponders were different in some way.
Another limitation in this study design is the selection of the survival data. Survival data for this study were taken from 2 large multicenter databases that were the most recent sources available to respondents at the time of the survey.11,,12 Since then, more recent outcome results have been published by the NICHD Neonatal Research Network.19 These results evaluate a cohort born between January 1993 and December 1994. In all birth weight categories between 501 g and 1500 g, mortality and morbidity decreased. Although the databases are primarily from academic centers and the questionnaire primarily targeted nonacademic physicians, recent data indicate that outcomes at nonacademic centers are comparable to academic centers in the very low birth weight population.20,,21 Most, but not all, recent studies report that survival of preterm infants continues to improve.6,,7,22,23 If more current data with improved survival had been used for comparison, the underestimation by the physicians surveyed would have even been more marked. In contrast, despite the increase in survival, handicap-free rates reported in recent years continue to be comparable to the reference rates used in this study.24–29
Uniform standards for the prediction and description of long-term impairment in children are elusive. Handicap rates for surviving preterm infants are measured and reported according to variable criteria, including type of impairment (sensory, motor, and intellectual), age at follow-up, and testing protocol. Although most investigators agree on what generally constitutes severe handicap, it is possible that the physicians' interpretations differ, accounting for part of the underestimation of handicap-free rates. No precise definition of severe handicap was given in the questionnaire because we wanted the physicians to give an overall estimate of any severely handicapping conditions. Pediatricians' responses may have been influenced by their experience with preterm infants with normal early physical and neurologic examinations, who manifested difficulties later, such as on school entry.30,,31
Physicians' personal beliefs regarding outcome have been shown to influence their practice policies, counseling messages, and use of therapies.32,,33 In a recent review of the ethical dilemmas of delivery room resuscitation, Stevenson and Goldworth34caution that “the likelihood of survival with or without injury … is influenced by the opinions and associated decisive behaviors of people who prejudge the biological incapacities before they are demonstrated.” Surveys of physicians' beliefs regarding the prognosis of preterm infants have been published using various formats. Wilson et al14 found that among obstetricians, family practitioners, and pediatricians, pediatricians were most accurate in their assessment of prognosis and that all physicians were more accurate in their prediction of mortality than morbidity. Among neonatologists surveyed in 1995, most counseled parents that mortality was 75% or greater at gestational ages of 24 weeks or less and up to 50% for gestational ages between 25 and 27 weeks.15 These beliefs regarding outcome coincided with a majority of the same respondents not willing to resuscitate infants <27 weeks of gestation.15 More than 90% of obstetricians surveyed in California viewed future quality of life as dismal for infants 24 weeks or less.16 Based on this knowledge, >90% of these obstetricians would call a pediatrician to the delivery of a 25-week infant and 46% would do so for a 22-week infant.16 These reports exemplify the connection between physicians' beliefs and their actions regarding care, regardless of the accuracy of their beliefs.
Reduction in infant mortality continues to occur with advances in neonatal care.20,,35 However, many question the aggressive management of extremely immature neonates who might have significant morbidity if they survive.22 Survival has been shown to improve with improved access to advanced perinatal services.9,36–40 Despite increased rates of maternal and fetal referral to a perinatal center, there are many preterm infants born in hospitals without advanced perinatal services40,,41underscoring the importance of the attitudes of both obstetricians and pediatricians regarding lifesaving therapeutic practices for potentially viable infants. Very low birth weight infants not transferred to a center with neonatal intensive care services have significantly higher mortality rates in the first 24 hours.18
The theory of reasoned action has been applied to many health-related behaviors.42 Based on this theory, physicians believe that certain behaviors will lead to specific outcomes. Their knowledge of the outcome modifies their behavior and determines their attitudes. Furthermore, the views of physicians' peers (subjective norms) when combined with their own attitudes, form the intention to perform a behavior. This theory can provide a mechanism to identify points at which behavior can be modified. Based on the findings of this survey, it is clear that physicians' knowledge of outcomes is associated with their willingness to institute potentially lifesaving therapies for the neonate. To optimize patient care, it is essential that knowledge and subjective norms be based on the most accurate information.
In summary, these data show that physicians involved in the care of preterm infants underestimate their survival and handicap-free rates. In this study, physicians' estimates of survival rates were compared with nationally published rates for infants born just before the era of widespread use of surfactant and antenatal steroids. The underestimation of survival rates at shorter gestations would have been even more pronounced if compared with survival data for infants treated with more recent technologies.43 In addition, physicians' knowledge may lag behind current statistics, and this may explain the underestimation of outcome reported here. Improving physicians' knowledge regarding outcome of preterm infants may result in more appropriate use of therapeutic interventions that may be lifesaving.
This work was supported in part by Grant AL 30701007 from the Alabama Chapter, March of Dimes Birth Defects Foundation; by the Agency for Health Care Policy; and by Research Department of Health and Human Services Contract 290-92-0055 Patient Outcome Research Team on Low Birthweight.
We thank Dr John Waterbor for his review and comments regarding the statistical analysis.
- Received April 20, 1999.
- Accepted September 22, 1999.
Reprint requests to (W.A.C.) University of Alabama at Birmingham, Department of Pediatrics, 525 New Hillman Building, Birmingham, AL 35233. E-mail:
- NICHD =
- National Institute of Child Health and Human Development •
- CI =
- confidence interval
- Robertson CMT,
- Hrynchyshyn GJ,
- Eitches PC,
- Pain KS
- O'Shea TM,
- Klinepeter KL,
- Goldstein DJ,
- Jackson BW,
- Dillard RG
- Lee K,
- Kim BI,
- Khoshnood B,
- et al.
- Richardson DK,
- Gray JE,
- Gortmaker SL,
- et al.
- ↵Haywood JL, Morse SB, Goldenberg RL, Bronstein J, Nelson KG, Carlo WA. Estimation of outcome and restriction of interventions in neonates. Pediatrics. 1998(2). URL: http://www.pediatrics.org/cgi/content/full/102/2/e6
- Hack M,
- Horbar J,
- Malloy M,
- Tyson J,
- Wright E,
- Wright L
- Weiss E,
- Martinez AM,
- Greeman H,
- Partridge JC
- ↵Aday L. General principles for formulating questions. In: Designing and Conducting Health Surveys—A Comprehensive Guide. San Francisco, CA: Jossey-Bass Inc; 1989:129–194
- The Vermont-Oxford Trials Network
- Hack M,
- Friedman H,
- Fanaroff AA
- Ling E,
- Battin M,
- Whitfield M
- Johnson A,
- Townshend P,
- Yudkin P,
- Bull D,
- Wilkinson A
- Bhushan V,
- Paneth N,
- Kiely J
- Higgins SS,
- Paul SM,
- Hardy CE,
- Ternullo-Retta C,
- Affonso D
- Guyer B,
- Strobino DM,
- Ventura SJ,
- et al.
- Thompson M,
- Khot A
- Cifuentes J,
- Bronstein J,
- Phibbs C,
- Phibbs R,
- Carlo WA
- ↵Ajzen I, Fishbein M. Belief, attitude, intention, and behavior: an introduction to theory and research. Reading, MA: Addison-Wesley Publishing Co; 1975:288–334
- Horbar JD,
- Wright EC,
- Onstad L,
- et al.
- Copyright © 2000 American Academy of Pediatrics