- Email: [email protected]
Population-Based Outcomes Data for Counseling at the Margin of Gestational Viability
ARTICLE IN PRESS THE JOURNAL OF PEDIATRICS • www.jpeds.com
ORIGINAL ARTICLES
Population-Based Outcomes Data for Counseling at the Margin of Gestational Viability Patrick Myers, MD1, Naomi Laventhal, MD2, Bree Andrews, MD, MPH3, Joanne Lagatta, MD4, and William Meadow, MD, PhD3 Objective To survey neonatologists as to how many use population-based outcomes data to counsel families before and after the birth of 22- to 25-week preterm infants.
Study design An anonymous online survey was distributed to 1022 neonatologists in the US. Questions addressed the use of population-based outcome data in prenatal and postnatal counseling. Results Ninety-one percent of neonatologists reported using population-based outcomes data for counseling. The National Institute of Child Health and Human Development Neonatal Research Network Outcomes Data is most commonly used (65%) with institutional databases (14.5%) the second choice. Most participants (89%) reported that these data influence their counseling, but it was less clear whether specific estimates of mortality and morbidity influenced families; 36% of neonatologist felt that these data have little or no impact on families. Seventyone percent reported that outcomes data estimates confirmed their own predictions, but among those who reported having their assumptions challenged, most had previously been overly pessimistic. Participants place a high value on gestational age and family preference in counseling; however, among neonatologists in high-volume centers, the presence of fetal complications was also reported to be an important factor. A large portion of respondents reported using prenatal population-based outcomes data in the neonatal intensive care unit. Conclusion Despite uncertainty about their value and impact, neonatologists use population-based outcomes data and provide specific estimates of survival and morbidity in consultation before and after extremely preterm birth. How best to integrate these data into comprehensive, family-centered counseling of infants at the margin of viability is an important area of further study. (J Pediatr 2016;■■:■■-■■). he incidence of mortality1 and major neonatal morbidity2 has improved for extremely preterm infants on a population level, but it remains difficult to predict the trajectory of each individual infant, particularly before birth. Uncertainty around individual postnatal trajectories of illness complicates counseling families about the risks and benefits of starting, continuing, withholding, or withdrawing life-sustaining interventions. Historically, gestational age has been the focus of how the lower limits of viability have been described and discussed. Multicenter collaborative efforts such as the Vermont Oxford Network2 and Pediatrix3 have led to user-friendly short- and longterm outcomes databases that can be used in prenatal discussions with families. The National Institute of Child Health and Human Development (NICHD) Neonatal Research Network: Extremely Preterm Birth Outcome Data4 (NICHD outcomes calculator) added neurodevelopmental follow up and a multifactorial approach to prognostication, highlighting the impact of other fetal attributes on outcomes for extremely preterm infants. Along with aggregations of institutional outcomes data, this body of work has created population-based outcomes data that provide data to neonatologists about possible neonatal outcomes. It is not known how and when neonatologists use these population-based outcomes data or which factors most influence how neonatologists frame discussions with families. Regional surveys5 have evaluated practice patterns and governing bodies have issued policy statements6-10 to guide practitioners. Prenatal counseling is done frequently, but with a wide range of variation across different centers.11 Less is known about how population-based outcomes data are used to frame conversations with families at risk of delivering at the margins of viability or in neonatal intensive care units (NICU). Against this background, we set out to survey neonatologists in the US to better understand the role of population-based outcome prediction data.
T
Methods Surveys (Appendix; available at www.jpeds.com) were sent to all listed neonatologists and neonatology fellows in 2 organizations from whom e-mail addresses were accessible during October of 2014: the American Academy of Pediatrics Section on Perinatal Pediatrics and the Organization of Neonatal-Perinatal Medicine Training
From the 1Feinberg School of Medicine, Northwestern University, Chicago, IL; 2C.S. Mott Children’s Hospital, University of Michigan, Ann Arbor, MI; 3The University of Chicago Comer Children’s Hospital, Chicago, IL; and 4 Medical College of Wisconsin, Milwaukee, WI The authors declare no conflicts of interest.
NICHD NICU
National Institute of Child Health and Human Development Neonatal intensive care units
0022-3476/$ - see front matter. © 2016 Elsevier Inc. All rights reserved. http://dx.doi.org10.1016/j.jpeds.2016.10.021
1 FLA 5.4.0 DTD ■ YMPD8731_proof ■ November 1, 2016
THE JOURNAL OF PEDIATRICS • www.jpeds.com Program Directors. This wide-net approach yielded a large number of e-mail addresses that were no longer operational but, owing to concerns about anonymity, we were unable to exclude the nonfunctional e-mail addresses before sending the survey. Potential participants were sent an email, given a short study description, assurance of anonymity, and the opportunity to decline. Each participant was asked to confirm their professional role as a neonatologist and participants were excluded if they did meet this screening criterion. The study was submitted and exempted by institutional review boards. In our survey, we defined population-based outcomes data as “aggregated sources of outcomes data (i.e. the NICHD Neonatal Research Network: Extremely Preterm Birth Outcomes Data (NICHD outcomes calculator), Pediatrix/Obstetrix Outcomes Data, Vermont Oxford Network, or institutional outcomes data).” This survey asked clinicians to focus their responses to infants at 22-25 completed weeks gestation. Qualtrics (Provo, Utah) was used to send the survey, send 2 reminders, collect responses, and provide anonymity. Demographic information was gathered, namely, years in practice, frequency of prenatal counseling, hospital characteristics, and specialty. Before the deployment of the full survey, multiple pilot surveys were sent to optimize both survey questions and skip logic flow. Pilot respondents helped to improve survey flow and to clarify questions and answers choices. Questions were divided into prenatal and postnatal categories and focused on the use of population-based outcomes calculators, perception of families’ response to empiric data, and the respondents’ priorities in counseling. For rank order questions, participants were given the option of ranking only those choices that they felt applied. Stata SE version 13 (StataCorp, College Station, Texas) was used to analyze the data. We used c2 tests or Fisher exact test (for cell size < 5) to compare the different characteristics among participants.
Results Of 7100 functional and nonfunctional e-mail addresses, there were 1291 emails that were opened and 1225 people (95%) completed the survey. Of the participants, 1022 (83%) confirmed that they were neonatologists and are the focus of this paper. Among website visitors, we had a 95% response rate, but our overall response rate was only 18%. Respondent demographics (Table) showed that the average respondent had worked for >20 years, practiced in a hospital with planned highrisk deliveries, could provide mechanical ventilation, and had more than one 22- to 25-week infant born per month. Use of Population-Based Outcomes Data The vast majority (99%) of surveyed neonatologists provided counseling to parents at risk of delivery between 22 and 25 weeks. Of the participating neonatologists, 91% used data to counsel families prenatally, with 80% using data 50% of the time or more and 30% reporting using data all of the time. The NICHD outcomes calculator4 was the most frequently used population-based outcomes data source (Figure 1).
Volume ■■
Table. Demographics Questions How many years have you been practicing since completing your training? Fellow <3 3-5 6-10 11-20 >20 How often do you counsel families of infants born at or before 25 completed weeks in the NICU, who might consider withdrawal of life-prolonging interventions? Less than once a year Few times a year Few times a month Few times a week How many infants are born between 22 and 25 completed weeks at the hospital where you most often provide counseling? None 1-2 every couple of years About once a month About twice a month More than twice per month What best describes the primary NICU where you most often provide counseling? No planned high-risk deliveries; cannot provide mechanical ventilation No planned high-risk deliveries; can provide mechanical ventilation Planned high-risk deliveries; can provide mechanical ventilation No deliveries; can provide mechanical ventilation
n
%
65 88 122 157 167 423
6 9 12 15 16 41
110 693 173 9
11 70 18 1
7 86 231 253 445
1 8 23 25 44
1
0
26
4
618
94
15
2
Institutionally derived data compiled for the participant’s own hospital system was second followed by almost equal numbers for Pediatrix/Obsterix3 and Vermont Oxford2 datasets. The majority (78%) of those who picked the“other”option and entered free text indicate that they used multiple sources to counsel prenatally. Impact of Population-Based Outcomes Data on Prenatal Counseling Of those neonatologists who consult data before counseling, 89% stated that population-based outcomes data had at least some impact on how they counseled. Despite consulting a data source before counseling, 11% of respondents felt populationbased outcomes data had “little” or “no” impact on how they counseled. Of the neonatologists who used population-based outcomes data, 80% said that they provide families “specific numerical estimates of adverse outcomes, such as survival or survival without neurologic impairment.”When asked how much impact data had on parent’s decisions about initiating or withholding resuscitative efforts in the delivery room, 36% stated that these estimates had “little” or “no” impact on parents. The majority felt that numerical estimates had at least some effect and 10% responded that data estimates had a major impact on families. Neonatologists who use population-based outcomes data were asked, “When you use a population-based outcomes data,
2
Myers et al FLA 5.4.0 DTD ■ YMPD8731_proof ■ November 1, 2016
ORIGINAL ARTICLES
2016
Figure 1. Population-based outcome data choice among neonatologists who reported using population-based outcomes data for prenatal consultation. VON, Vermont Oxford Network.
how often do you find that the estimates for morbidity and/or mortality are different than you expected?” Six percent of neonatologists felt that their estimates were different from a population-based outcomes data all or most of the time, and 71% responded that their estimate was different infrequently or never. When asked, “In cases in which the outcomes you find are unexpected, do you usually find that you overestimated or underestimated adverse outcomes?”, 21% reported underestimating the incidence of adverse outcomes and 40% said they overestimated adverse outcomes. Neonatologists were asked to rank order (Figure 2) only those factors that most “affect your recommendations about initiating or withholding resuscitative efforts in the delivery room.” Family preference, presence of major anomalies, gestational age, and fetal weight were the most ranked items. Gestational age was the most frequently the first choice, with almost 50% of all respondents ranking it first. The free text “other” choice was selected 2% of the time and most frequently indicated the use of postdelivery factors to make recommendations about initiating or withholding care. Only 4% of neonatologists counseled a few times a week or more, but they represented a unique group. They were twice as likely (38% vs 16%) to use their own institutional data (P = .004). They were also more likely to reference a populationbased outcomes dataset before counseling a family compared with neonatologists who counseled less frequently (P < .001). These higher volume counselors were also almost 3 times more likely (21% vs 8%; P = .017) to rank the “presence or absence of major fetal complications, anomalies,
infection, etc” as the most important factor when recommending initiation or withholding resuscitation efforts in the delivery room to families. Postnatal Use of Population-Based Outcomes Data Postnatal use of population-based outcomes data was reported by 57% of neonatologists to counsel families of infants in the NICU when considering withdrawal of life-prolonging interventions. The majority of postnatal counseling (55%) was with the NICHD4 calculator, which considers only factors known before birth and was not designed for postnatal use. Three groups of neonatologists were significantly more likely to use population-based outcomes data to counsel in the NICU: neonatologists who felt most strongly that population-based outcomes-derived data affected how they counseled (P = .001), neonatologists who stated that families were more likely to make decisions based on population-based outcomes-data derived data (P > .001), and more experienced neonatologists (P = .001).
Discussion Our large, national survey of US neonatologists specifically examined the role of population-based outcomes data for counseling at the margin of gestational viability. With the widespread availability of population-based outcomes data reporting outcomes for extremely preterm infants, it is not surprising that most neonatologists report using some sort of data to aid in prenatal counseling. However, our participants were not uniform in their perceived impact and value of these data for
Population-Based Outcomes Data for Counseling at the Margin of Gestational Viability FLA 5.4.0 DTD ■ YMPD8731_proof ■ November 1, 2016
3
THE JOURNAL OF PEDIATRICS • www.jpeds.com
Volume ■■
Figure 2. Most respondents ranked many choices as important indicating the influence of multiple factors when counseling. Family preference, major fetal complications, gestational age, and fetal weight were most often ranked in the top 3 or 5. Gestational age was most frequently the highest ranked choice.
expectant parents, particularly with regard to the provision of actual numeric estimates of mortality and morbidity. Grobman et al12 showed that care providers focused on “objective” facts, whereas families choose to focus on less objective measures such as “hope.” Mothers have also been shown to not focus on morbidity and mortality predictions as a central part of their decision making for delivery room resuscitation.13 The subgroup of neonatologists (11%) who use population-based outcomes data despite feeling that they have little impact on their prenatal counseling highlight the difficulty in integrating the complex and nuanced needs of families receiving prenatal counseling and the available resources to support clinicians who provide it. The impact and usefulness of providing specific numerical estimates of morbidity and mortality on families and neonatologists requires continued study. When asked to compare their intuitions about morbidity and mortality to data generated by a population-based outcomes data, most neonatologists said that their intuitions were similar to the data generated, but a minority reported systematically having to correct their assumptions. Neonatologists have been shown to have poor prognostic intuitions, with less than a 50% chance of determining an intubated infant’s mortality.14 This is consistent with previous findings that suggest that physicians’ pessimism regarding infants at the border of viability,15 especially when compared
with attitudes of families and survivors.16 Given this pessimism and the high frequency with which neonatologists in our study reported providing numeric prognostic estimates, our study suggests that whether or not specific estimates of mortality and morbidity are provided to families, populationbased outcomes data may have a valuable role in neonatologist’s preparation for prenatal consultation, to offset this bias. We found that a substantial proportion of neonatologists reporting use of population-based outcomes data for counseling families with infants after delivery in the NICU. The majority used a populations-based outcomes data tool that does not incorporate any information acquired after birth. Some may have only used the postnatal data built into some aggregate data sources, but our survey did not explore this issue. This is surprising; both short- and long-term outcomes quickly diverge from prenatal prognostication.17,18 Prenatal populationbased outcome prediction tools were not designed for use in infants in the NICU, and postnatal counseling data have been developed17,19 for the use in the NICU population. Perhaps the frequent use of prenatal population-based outcomes data in the NICU reflects a comfort level with or an ease of access to these data sources. The differences between those neonatologists who counsel most often and other neonatologists may be the result of several factors. NICUs large enough to generate reliable institutional
4
Myers et al FLA 5.4.0 DTD ■ YMPD8731_proof ■ November 1, 2016
ORIGINAL ARTICLES
2016 outcomes data are likely to require the most prenatal consultations. It may also reflect the knowledge that institutional data are more specific than data derived at a national level, and may offer institutional experience with specific congenital anomalies or fetal complications that are not captured by multicenter current outcomes databases. The consideration of major fetal complications also seems to reflect recent epidemiologic studies of patients with congenital anomalies, whose outcomes are often unfavorably influenced by prematurity.20-22 Our study has identifiable limitations. The complex interactions and conversations that happen between a family and the medical team providing prenatal counseling cannot be captured fully in a survey. It is likely that some survey answers reflect the best choice but not the whole choice that an actual neonatologist would make when counseling. The responses to our survey questions are the perceptions of the respondents, but need prospective investigation to validate the actual practice patterns because practice and perception can vary. Additionally, although we had more than 1000 responses, the actual response rate was low, introducing the possibility of selection bias. For families expecting an infant to be born at the border of gestational viability, neonatologists use population-based outcomes data as part of the complex and multifactorial counseling process. Population-based outcomes data provide a starting point for both neonatologists and families, but the best way to use epidemiologic data in individual counseling remains unclear. Further studies exploring the needs and desires of these families are needed. The use of prenatal counseling data in postnatal counseling and the lack of data on fetal complications or anomalies, suggests the need for continued development of appropriate prognostication tools. ■ Submitted for publication Jun 25, 2016; last revision received Sep 1, 2016; accepted Oct 5, 2016 Reprint requests: Patrick Myers MD, Ann & Robert H. Lurie Children’s Hospital of Chicago Section of Neonatology, 225 E Chicago Ave, Chicago, IL 60611. E-mail: [email protected]
References 1. Wilson-Costello D, Friedman H, Minich N, Fanaroff AA, Hack M. Improved survival rates with increased neurodevelopmental disability for extremely low birth weight infants in the 1990s. Pediatrics 2005;115:9971003. 2. Horbar JD, Carpenter JH, Badger GJ, Kenny MJ, Soll RF, Morrow KA, et al. Mortality and neonatal morbidity among infants 501 to 1500 grams from 2000 to 2009. Pediatrics 2012;129:1019-26. 3. Pediatrix Medical Group. Outcomes data - Pediatrix Medical Group. 2012 http://www.pediatrix.com/OutcomesData. Accessed July 24, 2015.
4. Tyson JE, Parikh NA, Langer J, Green C, Higgins RD. Intensive care for extreme prematurity–moving beyond gestational age. N Engl J Med 2008;358:1672-81. 5. Bastek TK, Richardson DK, Zupancic JA, Burns JP. Prenatal consultation practices at the border of viability: a regional survey. Pediatrics 2005;116:407-13. 6. Bell EF, Batton DG, Stark AR. Noninitiation or withdrawal of intensive care for high- risk newborns: in reply. Pediatrics 2007;119:1267-9. 7. MacDonald H. Perinatal care at the threshold of viability. Pediatrics 2002;110:1024-7. 8. Kattwinkel J, Perlman JM, Aziz K, Colby C, Fairchild K, Gallagher J, et al. Part 15: neonatal resuscitation: 2010 American Heart Association Guidelines for Cardiopulmonary Resuscitation and Emergency Cardiovascular Care. Circulation 2010;122:S909-19. 9. Ecker JL, Kaimal A, Mercer BM, Blackwell SC, deRegnier RA, Farrell RM, et al. American College of Obstetricians and Gynecologists. # 3: periviable birth. Am J Obstet Gynecol 2015;213:604-14. 10. Cummings J, Watterberg K, Eichenwald E, Poindexter B, Stewart DL, Aucott SW, et al. Antenatal counseling regarding resuscitation and intensive care before 25 weeks of gestation. Pediatrics 2015;136:588-95. 11. Mehrotra A, Lagatta J, Simpson P, Kim UO, Nugent M, Basir MA. Variations among US hospitals in counseling practices regarding prematurely born infants. J Perinatol 2013;33:509-13. 12. Grobman WA, Kavanaugh K, Moro T, deRegnier RA, Savage T. Providing advice to parents for women at acutely high risk of periviable delivery. Obstet Gynecol 2010;115:904. 13. Boss RD, Hutton N, Sulpar LJ, West AM, Donohue PK. Values parents apply to decision-making regarding delivery room resuscitation for highrisk newborns. Pediatrics 2008;122:583-9. 14. Meadow W, Lagatta J, Andrews B, Caldarelli L, Keiser A, Laporte J, et al. Just, in time: ethical implications of serial predictions of death and morbidity for ventilated premature infants. Pediatrics 2008;121:732-40. 15. Janvier A, Lantos J, Deschenes M, Couture E, Nadeau S, Barrington KJ. Caregivers attitudes for very premature infants: what if they knew? Acta Paediatr 2008;97:276-9. 16. Saigal S, Stoskopf BL, Feeny D, Furlong W, Burrows E, Rosenbaum PL, et al. Differences in preferences for neonatal outcomes among health care professionals, parents, and adolescents. JAMA 1999;281:1991-7. 17. Ambalavanan N, Carlo WA, Tyson JE, Langer JC, Walsh MC, Parikh NA, et al. Outcome trajectories in extremely preterm infants. Pediatrics 2012;130:e115-25. 18. Meadow W, Frain L, Ren Y, Lee G, Soneji S, Lantos J. Serial assessment of mortality in the neonatal intensive care unit by algorithm and intuition: certainty, uncertainty, and informed consent. Pediatrics 2002;109:87886. 19. Andrews B, Myers P, Lagatta J, Meadow W. A comparison of prenatal and postnatal models to predict outcomes at the border of viability. J Pediatr 2016;173:96-100. 20. Pappas A, Shankaran S, Hansen NI, Bell EF, Stoll BJ, Laptook AR, et al. Outcome of extremely low birth weight infants with congenital heart defects in the Eunice Kennedy Shriver NICHD Neonatal Research Network. Pediatr Cardiol 2012;33:1415. 21. Walden RV, Taylor SC, Hansen NI, Poole WK, Stoll BJ, Abuelo D, et al. Major congenital anomalies place extremely low birth weight infants at higher risk for poor growth and developmental outcomes. Pediatrics 2007;120:e1512-9. 22. Lynema S, Fifer CG, Laventhal NT. Perinatal decision making for preterm infants with congenital heart disease: determinable risk factors for mortality. Pediatr Cardiol 2016;1-8.
Population-Based Outcomes Data for Counseling at the Margin of Gestational Viability FLA 5.4.0 DTD ■ YMPD8731_proof ■ November 1, 2016
5
THE JOURNAL OF PEDIATRICS • www.jpeds.com
Volume ■■
Appendix. All survey questions with skip logic prompts. (Continues)
5.e1
Myers et al FLA 5.4.0 DTD ■ YMPD8731_proof ■ November 1, 2016
ORIGINAL ARTICLES
2016
Appendix. Continues.
Population-Based Outcomes Data for Counseling at the Margin of Gestational Viability FLA 5.4.0 DTD ■ YMPD8731_proof ■ November 1, 2016
5.e2
THE JOURNAL OF PEDIATRICS • www.jpeds.com
Volume ■■
Appendix. Continues.
5.e3
Myers et al FLA 5.4.0 DTD ■ YMPD8731_proof ■ November 1, 2016
ORIGINAL ARTICLES
2016
Appendix. Continued.
Population-Based Outcomes Data for Counseling at the Margin of Gestational Viability FLA 5.4.0 DTD ■ YMPD8731_proof ■ November 1, 2016
5.e4
ORIGINAL ARTICLES
Population-Based Outcomes Data for Counseling at the Margin of Gestational Viability Patrick Myers, MD1, Naomi Laventhal, MD2, Bree Andrews, MD, MPH3, Joanne Lagatta, MD4, and William Meadow, MD, PhD3 Objective To survey neonatologists as to how many use population-based outcomes data to counsel families before and after the birth of 22- to 25-week preterm infants.
Study design An anonymous online survey was distributed to 1022 neonatologists in the US. Questions addressed the use of population-based outcome data in prenatal and postnatal counseling. Results Ninety-one percent of neonatologists reported using population-based outcomes data for counseling. The National Institute of Child Health and Human Development Neonatal Research Network Outcomes Data is most commonly used (65%) with institutional databases (14.5%) the second choice. Most participants (89%) reported that these data influence their counseling, but it was less clear whether specific estimates of mortality and morbidity influenced families; 36% of neonatologist felt that these data have little or no impact on families. Seventyone percent reported that outcomes data estimates confirmed their own predictions, but among those who reported having their assumptions challenged, most had previously been overly pessimistic. Participants place a high value on gestational age and family preference in counseling; however, among neonatologists in high-volume centers, the presence of fetal complications was also reported to be an important factor. A large portion of respondents reported using prenatal population-based outcomes data in the neonatal intensive care unit. Conclusion Despite uncertainty about their value and impact, neonatologists use population-based outcomes data and provide specific estimates of survival and morbidity in consultation before and after extremely preterm birth. How best to integrate these data into comprehensive, family-centered counseling of infants at the margin of viability is an important area of further study. (J Pediatr 2016;■■:■■-■■). he incidence of mortality1 and major neonatal morbidity2 has improved for extremely preterm infants on a population level, but it remains difficult to predict the trajectory of each individual infant, particularly before birth. Uncertainty around individual postnatal trajectories of illness complicates counseling families about the risks and benefits of starting, continuing, withholding, or withdrawing life-sustaining interventions. Historically, gestational age has been the focus of how the lower limits of viability have been described and discussed. Multicenter collaborative efforts such as the Vermont Oxford Network2 and Pediatrix3 have led to user-friendly short- and longterm outcomes databases that can be used in prenatal discussions with families. The National Institute of Child Health and Human Development (NICHD) Neonatal Research Network: Extremely Preterm Birth Outcome Data4 (NICHD outcomes calculator) added neurodevelopmental follow up and a multifactorial approach to prognostication, highlighting the impact of other fetal attributes on outcomes for extremely preterm infants. Along with aggregations of institutional outcomes data, this body of work has created population-based outcomes data that provide data to neonatologists about possible neonatal outcomes. It is not known how and when neonatologists use these population-based outcomes data or which factors most influence how neonatologists frame discussions with families. Regional surveys5 have evaluated practice patterns and governing bodies have issued policy statements6-10 to guide practitioners. Prenatal counseling is done frequently, but with a wide range of variation across different centers.11 Less is known about how population-based outcomes data are used to frame conversations with families at risk of delivering at the margins of viability or in neonatal intensive care units (NICU). Against this background, we set out to survey neonatologists in the US to better understand the role of population-based outcome prediction data.
T
Methods Surveys (Appendix; available at www.jpeds.com) were sent to all listed neonatologists and neonatology fellows in 2 organizations from whom e-mail addresses were accessible during October of 2014: the American Academy of Pediatrics Section on Perinatal Pediatrics and the Organization of Neonatal-Perinatal Medicine Training
From the 1Feinberg School of Medicine, Northwestern University, Chicago, IL; 2C.S. Mott Children’s Hospital, University of Michigan, Ann Arbor, MI; 3The University of Chicago Comer Children’s Hospital, Chicago, IL; and 4 Medical College of Wisconsin, Milwaukee, WI The authors declare no conflicts of interest.
NICHD NICU
National Institute of Child Health and Human Development Neonatal intensive care units
0022-3476/$ - see front matter. © 2016 Elsevier Inc. All rights reserved. http://dx.doi.org10.1016/j.jpeds.2016.10.021
1 FLA 5.4.0 DTD ■ YMPD8731_proof ■ November 1, 2016
THE JOURNAL OF PEDIATRICS • www.jpeds.com Program Directors. This wide-net approach yielded a large number of e-mail addresses that were no longer operational but, owing to concerns about anonymity, we were unable to exclude the nonfunctional e-mail addresses before sending the survey. Potential participants were sent an email, given a short study description, assurance of anonymity, and the opportunity to decline. Each participant was asked to confirm their professional role as a neonatologist and participants were excluded if they did meet this screening criterion. The study was submitted and exempted by institutional review boards. In our survey, we defined population-based outcomes data as “aggregated sources of outcomes data (i.e. the NICHD Neonatal Research Network: Extremely Preterm Birth Outcomes Data (NICHD outcomes calculator), Pediatrix/Obstetrix Outcomes Data, Vermont Oxford Network, or institutional outcomes data).” This survey asked clinicians to focus their responses to infants at 22-25 completed weeks gestation. Qualtrics (Provo, Utah) was used to send the survey, send 2 reminders, collect responses, and provide anonymity. Demographic information was gathered, namely, years in practice, frequency of prenatal counseling, hospital characteristics, and specialty. Before the deployment of the full survey, multiple pilot surveys were sent to optimize both survey questions and skip logic flow. Pilot respondents helped to improve survey flow and to clarify questions and answers choices. Questions were divided into prenatal and postnatal categories and focused on the use of population-based outcomes calculators, perception of families’ response to empiric data, and the respondents’ priorities in counseling. For rank order questions, participants were given the option of ranking only those choices that they felt applied. Stata SE version 13 (StataCorp, College Station, Texas) was used to analyze the data. We used c2 tests or Fisher exact test (for cell size < 5) to compare the different characteristics among participants.
Results Of 7100 functional and nonfunctional e-mail addresses, there were 1291 emails that were opened and 1225 people (95%) completed the survey. Of the participants, 1022 (83%) confirmed that they were neonatologists and are the focus of this paper. Among website visitors, we had a 95% response rate, but our overall response rate was only 18%. Respondent demographics (Table) showed that the average respondent had worked for >20 years, practiced in a hospital with planned highrisk deliveries, could provide mechanical ventilation, and had more than one 22- to 25-week infant born per month. Use of Population-Based Outcomes Data The vast majority (99%) of surveyed neonatologists provided counseling to parents at risk of delivery between 22 and 25 weeks. Of the participating neonatologists, 91% used data to counsel families prenatally, with 80% using data 50% of the time or more and 30% reporting using data all of the time. The NICHD outcomes calculator4 was the most frequently used population-based outcomes data source (Figure 1).
Volume ■■
Table. Demographics Questions How many years have you been practicing since completing your training? Fellow <3 3-5 6-10 11-20 >20 How often do you counsel families of infants born at or before 25 completed weeks in the NICU, who might consider withdrawal of life-prolonging interventions? Less than once a year Few times a year Few times a month Few times a week How many infants are born between 22 and 25 completed weeks at the hospital where you most often provide counseling? None 1-2 every couple of years About once a month About twice a month More than twice per month What best describes the primary NICU where you most often provide counseling? No planned high-risk deliveries; cannot provide mechanical ventilation No planned high-risk deliveries; can provide mechanical ventilation Planned high-risk deliveries; can provide mechanical ventilation No deliveries; can provide mechanical ventilation
n
%
65 88 122 157 167 423
6 9 12 15 16 41
110 693 173 9
11 70 18 1
7 86 231 253 445
1 8 23 25 44
1
0
26
4
618
94
15
2
Institutionally derived data compiled for the participant’s own hospital system was second followed by almost equal numbers for Pediatrix/Obsterix3 and Vermont Oxford2 datasets. The majority (78%) of those who picked the“other”option and entered free text indicate that they used multiple sources to counsel prenatally. Impact of Population-Based Outcomes Data on Prenatal Counseling Of those neonatologists who consult data before counseling, 89% stated that population-based outcomes data had at least some impact on how they counseled. Despite consulting a data source before counseling, 11% of respondents felt populationbased outcomes data had “little” or “no” impact on how they counseled. Of the neonatologists who used population-based outcomes data, 80% said that they provide families “specific numerical estimates of adverse outcomes, such as survival or survival without neurologic impairment.”When asked how much impact data had on parent’s decisions about initiating or withholding resuscitative efforts in the delivery room, 36% stated that these estimates had “little” or “no” impact on parents. The majority felt that numerical estimates had at least some effect and 10% responded that data estimates had a major impact on families. Neonatologists who use population-based outcomes data were asked, “When you use a population-based outcomes data,
2
Myers et al FLA 5.4.0 DTD ■ YMPD8731_proof ■ November 1, 2016
ORIGINAL ARTICLES
2016
Figure 1. Population-based outcome data choice among neonatologists who reported using population-based outcomes data for prenatal consultation. VON, Vermont Oxford Network.
how often do you find that the estimates for morbidity and/or mortality are different than you expected?” Six percent of neonatologists felt that their estimates were different from a population-based outcomes data all or most of the time, and 71% responded that their estimate was different infrequently or never. When asked, “In cases in which the outcomes you find are unexpected, do you usually find that you overestimated or underestimated adverse outcomes?”, 21% reported underestimating the incidence of adverse outcomes and 40% said they overestimated adverse outcomes. Neonatologists were asked to rank order (Figure 2) only those factors that most “affect your recommendations about initiating or withholding resuscitative efforts in the delivery room.” Family preference, presence of major anomalies, gestational age, and fetal weight were the most ranked items. Gestational age was the most frequently the first choice, with almost 50% of all respondents ranking it first. The free text “other” choice was selected 2% of the time and most frequently indicated the use of postdelivery factors to make recommendations about initiating or withholding care. Only 4% of neonatologists counseled a few times a week or more, but they represented a unique group. They were twice as likely (38% vs 16%) to use their own institutional data (P = .004). They were also more likely to reference a populationbased outcomes dataset before counseling a family compared with neonatologists who counseled less frequently (P < .001). These higher volume counselors were also almost 3 times more likely (21% vs 8%; P = .017) to rank the “presence or absence of major fetal complications, anomalies,
infection, etc” as the most important factor when recommending initiation or withholding resuscitation efforts in the delivery room to families. Postnatal Use of Population-Based Outcomes Data Postnatal use of population-based outcomes data was reported by 57% of neonatologists to counsel families of infants in the NICU when considering withdrawal of life-prolonging interventions. The majority of postnatal counseling (55%) was with the NICHD4 calculator, which considers only factors known before birth and was not designed for postnatal use. Three groups of neonatologists were significantly more likely to use population-based outcomes data to counsel in the NICU: neonatologists who felt most strongly that population-based outcomes-derived data affected how they counseled (P = .001), neonatologists who stated that families were more likely to make decisions based on population-based outcomes-data derived data (P > .001), and more experienced neonatologists (P = .001).
Discussion Our large, national survey of US neonatologists specifically examined the role of population-based outcomes data for counseling at the margin of gestational viability. With the widespread availability of population-based outcomes data reporting outcomes for extremely preterm infants, it is not surprising that most neonatologists report using some sort of data to aid in prenatal counseling. However, our participants were not uniform in their perceived impact and value of these data for
Population-Based Outcomes Data for Counseling at the Margin of Gestational Viability FLA 5.4.0 DTD ■ YMPD8731_proof ■ November 1, 2016
3
THE JOURNAL OF PEDIATRICS • www.jpeds.com
Volume ■■
Figure 2. Most respondents ranked many choices as important indicating the influence of multiple factors when counseling. Family preference, major fetal complications, gestational age, and fetal weight were most often ranked in the top 3 or 5. Gestational age was most frequently the highest ranked choice.
expectant parents, particularly with regard to the provision of actual numeric estimates of mortality and morbidity. Grobman et al12 showed that care providers focused on “objective” facts, whereas families choose to focus on less objective measures such as “hope.” Mothers have also been shown to not focus on morbidity and mortality predictions as a central part of their decision making for delivery room resuscitation.13 The subgroup of neonatologists (11%) who use population-based outcomes data despite feeling that they have little impact on their prenatal counseling highlight the difficulty in integrating the complex and nuanced needs of families receiving prenatal counseling and the available resources to support clinicians who provide it. The impact and usefulness of providing specific numerical estimates of morbidity and mortality on families and neonatologists requires continued study. When asked to compare their intuitions about morbidity and mortality to data generated by a population-based outcomes data, most neonatologists said that their intuitions were similar to the data generated, but a minority reported systematically having to correct their assumptions. Neonatologists have been shown to have poor prognostic intuitions, with less than a 50% chance of determining an intubated infant’s mortality.14 This is consistent with previous findings that suggest that physicians’ pessimism regarding infants at the border of viability,15 especially when compared
with attitudes of families and survivors.16 Given this pessimism and the high frequency with which neonatologists in our study reported providing numeric prognostic estimates, our study suggests that whether or not specific estimates of mortality and morbidity are provided to families, populationbased outcomes data may have a valuable role in neonatologist’s preparation for prenatal consultation, to offset this bias. We found that a substantial proportion of neonatologists reporting use of population-based outcomes data for counseling families with infants after delivery in the NICU. The majority used a populations-based outcomes data tool that does not incorporate any information acquired after birth. Some may have only used the postnatal data built into some aggregate data sources, but our survey did not explore this issue. This is surprising; both short- and long-term outcomes quickly diverge from prenatal prognostication.17,18 Prenatal populationbased outcome prediction tools were not designed for use in infants in the NICU, and postnatal counseling data have been developed17,19 for the use in the NICU population. Perhaps the frequent use of prenatal population-based outcomes data in the NICU reflects a comfort level with or an ease of access to these data sources. The differences between those neonatologists who counsel most often and other neonatologists may be the result of several factors. NICUs large enough to generate reliable institutional
4
Myers et al FLA 5.4.0 DTD ■ YMPD8731_proof ■ November 1, 2016
ORIGINAL ARTICLES
2016 outcomes data are likely to require the most prenatal consultations. It may also reflect the knowledge that institutional data are more specific than data derived at a national level, and may offer institutional experience with specific congenital anomalies or fetal complications that are not captured by multicenter current outcomes databases. The consideration of major fetal complications also seems to reflect recent epidemiologic studies of patients with congenital anomalies, whose outcomes are often unfavorably influenced by prematurity.20-22 Our study has identifiable limitations. The complex interactions and conversations that happen between a family and the medical team providing prenatal counseling cannot be captured fully in a survey. It is likely that some survey answers reflect the best choice but not the whole choice that an actual neonatologist would make when counseling. The responses to our survey questions are the perceptions of the respondents, but need prospective investigation to validate the actual practice patterns because practice and perception can vary. Additionally, although we had more than 1000 responses, the actual response rate was low, introducing the possibility of selection bias. For families expecting an infant to be born at the border of gestational viability, neonatologists use population-based outcomes data as part of the complex and multifactorial counseling process. Population-based outcomes data provide a starting point for both neonatologists and families, but the best way to use epidemiologic data in individual counseling remains unclear. Further studies exploring the needs and desires of these families are needed. The use of prenatal counseling data in postnatal counseling and the lack of data on fetal complications or anomalies, suggests the need for continued development of appropriate prognostication tools. ■ Submitted for publication Jun 25, 2016; last revision received Sep 1, 2016; accepted Oct 5, 2016 Reprint requests: Patrick Myers MD, Ann & Robert H. Lurie Children’s Hospital of Chicago Section of Neonatology, 225 E Chicago Ave, Chicago, IL 60611. E-mail: [email protected]
References 1. Wilson-Costello D, Friedman H, Minich N, Fanaroff AA, Hack M. Improved survival rates with increased neurodevelopmental disability for extremely low birth weight infants in the 1990s. Pediatrics 2005;115:9971003. 2. Horbar JD, Carpenter JH, Badger GJ, Kenny MJ, Soll RF, Morrow KA, et al. Mortality and neonatal morbidity among infants 501 to 1500 grams from 2000 to 2009. Pediatrics 2012;129:1019-26. 3. Pediatrix Medical Group. Outcomes data - Pediatrix Medical Group. 2012 http://www.pediatrix.com/OutcomesData. Accessed July 24, 2015.
4. Tyson JE, Parikh NA, Langer J, Green C, Higgins RD. Intensive care for extreme prematurity–moving beyond gestational age. N Engl J Med 2008;358:1672-81. 5. Bastek TK, Richardson DK, Zupancic JA, Burns JP. Prenatal consultation practices at the border of viability: a regional survey. Pediatrics 2005;116:407-13. 6. Bell EF, Batton DG, Stark AR. Noninitiation or withdrawal of intensive care for high- risk newborns: in reply. Pediatrics 2007;119:1267-9. 7. MacDonald H. Perinatal care at the threshold of viability. Pediatrics 2002;110:1024-7. 8. Kattwinkel J, Perlman JM, Aziz K, Colby C, Fairchild K, Gallagher J, et al. Part 15: neonatal resuscitation: 2010 American Heart Association Guidelines for Cardiopulmonary Resuscitation and Emergency Cardiovascular Care. Circulation 2010;122:S909-19. 9. Ecker JL, Kaimal A, Mercer BM, Blackwell SC, deRegnier RA, Farrell RM, et al. American College of Obstetricians and Gynecologists. # 3: periviable birth. Am J Obstet Gynecol 2015;213:604-14. 10. Cummings J, Watterberg K, Eichenwald E, Poindexter B, Stewart DL, Aucott SW, et al. Antenatal counseling regarding resuscitation and intensive care before 25 weeks of gestation. Pediatrics 2015;136:588-95. 11. Mehrotra A, Lagatta J, Simpson P, Kim UO, Nugent M, Basir MA. Variations among US hospitals in counseling practices regarding prematurely born infants. J Perinatol 2013;33:509-13. 12. Grobman WA, Kavanaugh K, Moro T, deRegnier RA, Savage T. Providing advice to parents for women at acutely high risk of periviable delivery. Obstet Gynecol 2010;115:904. 13. Boss RD, Hutton N, Sulpar LJ, West AM, Donohue PK. Values parents apply to decision-making regarding delivery room resuscitation for highrisk newborns. Pediatrics 2008;122:583-9. 14. Meadow W, Lagatta J, Andrews B, Caldarelli L, Keiser A, Laporte J, et al. Just, in time: ethical implications of serial predictions of death and morbidity for ventilated premature infants. Pediatrics 2008;121:732-40. 15. Janvier A, Lantos J, Deschenes M, Couture E, Nadeau S, Barrington KJ. Caregivers attitudes for very premature infants: what if they knew? Acta Paediatr 2008;97:276-9. 16. Saigal S, Stoskopf BL, Feeny D, Furlong W, Burrows E, Rosenbaum PL, et al. Differences in preferences for neonatal outcomes among health care professionals, parents, and adolescents. JAMA 1999;281:1991-7. 17. Ambalavanan N, Carlo WA, Tyson JE, Langer JC, Walsh MC, Parikh NA, et al. Outcome trajectories in extremely preterm infants. Pediatrics 2012;130:e115-25. 18. Meadow W, Frain L, Ren Y, Lee G, Soneji S, Lantos J. Serial assessment of mortality in the neonatal intensive care unit by algorithm and intuition: certainty, uncertainty, and informed consent. Pediatrics 2002;109:87886. 19. Andrews B, Myers P, Lagatta J, Meadow W. A comparison of prenatal and postnatal models to predict outcomes at the border of viability. J Pediatr 2016;173:96-100. 20. Pappas A, Shankaran S, Hansen NI, Bell EF, Stoll BJ, Laptook AR, et al. Outcome of extremely low birth weight infants with congenital heart defects in the Eunice Kennedy Shriver NICHD Neonatal Research Network. Pediatr Cardiol 2012;33:1415. 21. Walden RV, Taylor SC, Hansen NI, Poole WK, Stoll BJ, Abuelo D, et al. Major congenital anomalies place extremely low birth weight infants at higher risk for poor growth and developmental outcomes. Pediatrics 2007;120:e1512-9. 22. Lynema S, Fifer CG, Laventhal NT. Perinatal decision making for preterm infants with congenital heart disease: determinable risk factors for mortality. Pediatr Cardiol 2016;1-8.
Population-Based Outcomes Data for Counseling at the Margin of Gestational Viability FLA 5.4.0 DTD ■ YMPD8731_proof ■ November 1, 2016
5
THE JOURNAL OF PEDIATRICS • www.jpeds.com
Volume ■■
Appendix. All survey questions with skip logic prompts. (Continues)
5.e1
Myers et al FLA 5.4.0 DTD ■ YMPD8731_proof ■ November 1, 2016
ORIGINAL ARTICLES
2016
Appendix. Continues.
Population-Based Outcomes Data for Counseling at the Margin of Gestational Viability FLA 5.4.0 DTD ■ YMPD8731_proof ■ November 1, 2016
5.e2
THE JOURNAL OF PEDIATRICS • www.jpeds.com
Volume ■■
Appendix. Continues.
5.e3
Myers et al FLA 5.4.0 DTD ■ YMPD8731_proof ■ November 1, 2016
ORIGINAL ARTICLES
2016
Appendix. Continued.
Population-Based Outcomes Data for Counseling at the Margin of Gestational Viability FLA 5.4.0 DTD ■ YMPD8731_proof ■ November 1, 2016
5.e4