The Center-Effect on Outcomes for Infants Born at Less than 25 Weeks

Abstract

Marked variation exists in the care of infants born at <25 weeks’ gestation. The center or location where a fetus or infant is cared for influences outcomes at very early gestational ages. Understanding this “center-effect,” including characteristics associated with centers that have high survival of births at <25 weeks’ gestation, may inform future studies and guide care practices to improve outcomes. This review focuses on the impact that the location or center of birth has on survival and other important outcomes for infants born at <25 weeks gestation. We review potential sources of variation in care practices and other factors that might explain the “center-effect.”

Introduction

Since the 1950s, survival for infants born at <25 weeks’ gestation has steadily improved.¹ In many developed centers around the world, infants at these gestational ages are offered resuscitation and are cared for in the neonatal intensive care unit (NICU). The outcomes for these infants will depend on the antepartum and intrapartum care of the mother and fetus and the postnatal care of the infant. In considering the impact of center and location of birth, we initially discuss potential factors that might account for the effect that a center of birth can have on survival and other important neonatal outcomes (e.g., “center effect”) and then discuss specific obstetric and neonatal care practices that vary among centers and may explain differences in outcomes (Figure 1). We focus on factors that relate to infant survival, given its importance to families and clinicians, although we acknowledge the importance of many other outcomes that might be influenced by the center or location in which an infant is born and cared for.

Figure 1. Selected Care Practices and Other Factors That May Explain the “Center Effect.”.

Figure shows key obstetric and neonatal care practices, as well as other factors, that may explain differences in outcomes among centers. Please refer to the text for additional details regarding factors.

Variation in Neonatal Outcomes

Large magnitudes of center variation have been shown for a wide variety of neonatal outcomes, including survival and other morbidities. In a study of 756 US NICUs that were members of the Vermont Oxford Network (VON), adverse outcome rates for NICUs in the lowest decile of incidence for death or serious morbidity were 49 to 83% of the incidences for NICUs in the highest decile of adverse outcomes. The incidence rates of late onset infection, necrotizing enterocolitis, and severe retinopathy of prematurity were half as common (Table 1).² While adverse outcome rates improved over time, this study highlights the magnitude of variation in adverse outcomes within a single country, after accounting for case mix; differences in outcomes among units in 2014 were similar in magnitude to improvements in median performance of outcomes from 2005 to 2014.² Similar variations have been reported among different developed countries by the International Network for Evaluating Outcomes (iNEO) of Neonates. In a study of mortality among extremely preterm infants in neonatal networks of developed countries from 2007 through 2013, marked variations were observed in survival rates among networks, particularly at 24 weeks’ gestation;³ survival to hospital discharge rates ranged from 35% in Spain to 84% in Japan. Similar differences were observed among infants born at 25 to 27 weeks’ gestation, although of less absolute magnitude in differences as survival rates increased. Another study by the iNEO network evaluated death or severe intraventricular hemorrhage (IVH), periventricular leukomalacia (PVL), bronchopulmonary dysplasia (BPD), or treated retinopathy of prematurity (ROP) among 58,004 infants born at <1500 g at 24 to 31 weeks’ gestation and found similar variations, with the lowest rates of adverse outcomes in Switzerland (standardized ratio of 0.77; 99% CI 0.69–0.87) and highest rates in the United Kingdom (standardized ratio of 1.16; 95% CI 1.11–1.21), relative to the overall mortality rate.⁴

Table 1.

Mortality and Common Morbidities of the Top and Bottom Deciles of NICUs in the Vermont Oxford Network (VON) in 2014.

Outcome	10th percentile	90th percentile	Ratio of 10th to 90th percentile	Ratio of 50th percentile between 2014 and 2005
Mortality	9.9%	11.9%	0.83	0.78
Chronic Lung Disease (CLD)	21.6%	36.0%	0.60	0.90
Late-onset Infections	7.0%	13.6%	0.51	0.47
Necrotizing Enterocolitis (NEC)	3.7%	6.9%	0.54	0.72
Severe Intraventricular Hemorrhage (IVH)	7.2%	8.8%	0.82	0.84
Severe retinopathy of prematurity (ROP)	4.2%	8.6%	0.49	0.63

Percentiles reflect the distribution of outcomes among US NICUs. Risk adjusted outcome rates for infants with a birth weight of 501 to 1500 grams are shown in the 2^nd and 3^rd column. Data from Horbar et al.²

Desired and Undesired Variation in Care

In addition to variations in outcomes, wide variations in neonatal care were reported. As espoused by Sackett et al. in 1996, the goal of evidence-based care is to integrate individual clinical expertise with the best external evidence.⁵ This will require judgement and consideration of an individual family’s preferences and infant’s needs. Therefore, some variation in care should not only be expected, but perhaps, desired to avoid “cookbook” medicine, which may not allow for individualized care. Conversely, standardization of care through the development and application of protocols and guidelines has been shown to improve outcomes by removing undesired variation.⁶ Such standardization of care protocols, such as the use of order sets⁷ and checklists,⁸ may improve the quality of care, making it more safe, effective and reliable. In addition, reductions in measures of variation of care may reflect efforts to improve care and outcomes. These factors should be considered, when considering the degree of variability in neonatal care practices among centers, and how these practices may relate to the “center effect” on outcomes.

Quality of Care

In the EPICE study of private and public NICUs in 11 European countries, 17 practices or interventions were considered, and 5 were noted to be practices with high levels of evidence that were associated with lower rates of mortality and short-term morbidity, including delivery in units with appropriate neonatal intensive care services, administration of antenatal corticosteroids, hypothermia prevention, surfactant replacement therapy and breastfeeding and breast milk use.⁹ Four of these five practices were evaluated in 19 regions from 11 European countries for infants before 28 weeks’ gestation, and only 58% of infants received all four evidence-based practices, while most infants received at least one evidence-based practice (ranging from 74% for admission temperature ≥36.0° C to 89% for receipt of antenatal steroids).

Breastfeeding and breast milk use were not included in the evaluation, due to difficulties in measurement. After adjustment for potential confounding factors, the four evidence-based practices were associated with lower risks of in-hospital mortality or severe morbidity. The study reported that 18.3% of deaths in the non-evidence-based care group would have been prevented if 90% of infants received full evidence-based care. This study highlights the importance of considering not just a single intervention, but a package of multiple care practices, in evaluating the impact of quality of care among centers on neonatal outcomes.

In addition, the Baby-MONITOR effort involved weighing the importance of processes (e.g., care practices) and outcome measures as part of the development of a composite indicator of NICU quality.¹⁰ The Baby-MONITOR composite measure included receipt of antenatal steroids, hypothermia on admission, discharge on human milk feeding, and timely eye exam, as the 4 process measures, with antenatal steroid administration carrying the highest weight. This tool has been used to assess the overall quality of care at a given center, although centers may vary in performance of individual components (higher in one domain and lower in another).

Additionally, assessment of reports by centers with high survival at very early gestational ages (e.g., 22 weeks) may provide insight into practices important for survival. Single centers identified from a recent systematic review by Backes et al. that included estimates of survival for infants at 22 weeks’ gestation from 2010 onward are summarized in Table 2.¹¹ The data include outcomes from three centers with reported survival at 22 weeks’ gestation of >50%, among the highest reported survival rates among centers in the world: University of Iowa Hospitals and Clinics in the US,¹² University of Cologne in Germany,¹³ and University Children’s Hospital in Uppsala, Sweden.¹⁴ Care practices emphasized in the discussions of reports from these centers that may explain the high survival rates include the common provision of active treatment (resuscitation), high rates of antenatal steroid use, high rates of prenatal care, cesarean delivery, delayed cord clamping, and neonatologist presence in the delivery room. Additional articles in this series will focus on specific practices among these centers.

Table 2.

Characteristics and Survival of Reports from Individual Centers

Center (Country)ref	Iowa (USA)12	Cologne (Germany)13	Uppsala (Sweden)14	Nationwide (USA)14
Years	2006–2015	2010–2014	2006–2015	2006–2015
Sample size	255	106	40	16
Denominator	Live births	Live births and stillbirths	Live births (subset with active treatment)	Live births (subset with active treatment)
Inclusion	22–25 weeks	22–23 weeks	22 weeks	22 weeks
Exclusion	Congenital anomalies, death in delivery room, parental request for palliation	Received palliative care	Elective terminations, extramural deliveries, major congenital or chromosomal anomalies.	Elective terminations, extramural deliveries, major congenital or chromosomal anomalies.
Assessment of survival	To hospital discharge	To hospital discharge	1 year	1 year
Survival by Gestational age
22	14/20 (70%)	17/28 (61%)	21/40 (53%)	3/16 (19%)
23	41/50 (82%)	41/58 (71%)
24	70/79 (89%)
25	89/99 (90%)
Care practices
Active treatment	20/22 (91%) at 22 weeks; 50/52 at 23 weeks (96%)	28/45 (62%) at 22 weeks; 58/61 (95%) at 23 weeks	40/40 (100%)	16/72 (22%)
Any prenatal care	65/70 (93%) 169/178 (95%)
Any antenatal steroids	64/70 (91%) at 22–23 weeks; 166/173 (96%) at 24–25 weeks	26/28 (93%) at 22 weeks; 52/58 (90%) at 23 weeks	34/40 (85%) at 22 weeksa	4/16 (25%)a
Cesarean delivery	22/70 (31%) at 22–23 weeks; 123/178 (69%) At 24–25 weeks	21/28 (75%) at 22 weeks; 49/58 (84%) at 23 weeks	0/40 (0%) at 22 weeks	4/16 (31%) at 22 weeks
Surfactant use	70/70 (100%) at 22–23 weeks; 173/175 (99%) at 24–25 weeks	LISA 21/28 (75%) at 22 weeks; 42/58 (72%) at 23 weeks	40/40 (100%)b	9/16 (56%)b
Selected Outcomes
Tracheostomy	1/42 (2%) at 22–23 weeks; 3/125 (2%) at 24–25 weeks
Need for ventriculoperitoneal shunt	3/42 (7%) at 22–23 weeks; 3/121 (3%) at 24–25 weeks	4/28 (14%) at 22 weeks; 5/58 (9%) at 23 weeks
Laser or anti-VEGF for retinopathy of prematurity	4/45 (9%) at 22–23 weeks; 14/114 (12%) at 24–25 weeks)c	3/28 (11%) at 22 weeks; 1/58 (2%) at 23 weeks	10/21 (48%) among survivors	5/6 (67%) among survivors
Practices noted in report	High antenatal steroid use. High rate of prenatal care.	Antenatal steroid use, cesarean delivery, delayed cord clamping, lateral positioning of infant, stepwise increase in PEEP, less invasive surfactant administration	Antenatal steroid use, neonatologist presence in delivery room

Table includes all studies reported outcomes by single identifiable centers with years spanning 2010 and beyond that reported survival for 22 weeks’ gestation in systematic review Backes et al.¹¹

VEGF: vascular endothelial growth factor; LISA: less invasive surfactant application; PEEP: positive end expiratory pressure.

^a2 or more doses

^bIn delivery room

^cLaser only

These and other key obstetric and neonatal practices are highlighted in Figure 1 and are discussed in additional detail below. Of note, this is not intended to be a comprehensive list of care practices that might explain the “center effect,” but rather a focused list of key practices, based on the previously mentioned studies, that may explain what important practices are associated with improved survival at <25 weeks’ gestation. As shown in Table 2, even among centers with high rates of survival at 22 weeks’ gestation, variability in specific practices, such as cesarean delivery, exist. Therefore, multiple factors, as noted in the EPICE cohort study,⁹ rather than any single practice, are likely to explain some of the center-effects associated with the high survival rates for these high-performing centers. In addition, issues related to measurement and reporting of outcomes and other factors, as discussed in the next sections and highlighted in Figure 1, should be considered.

National and regional factors

As noted in a commentary to one of the aforementioned publications, many factors may influence variations in outcomes of a particular country, including economic, organizational, cultural and social differences, that can also influence variation in outcomes within centers in individual countries.¹⁵ Even in countries in similar regions, such as Scandinavia, approaches to treatment at 22 to 23 weeks’ gestation can differ markedly, with approaches developed after considering factors such as public sentiment, professional preferences, reported outcomes, philosophical factors, and considerations of cost and cost-effectiveness.¹⁶ The context of these national and regional factors should be considered when comparing outcomes between geographically separate centers.

Segregation and case mix of patients

In addition to national and geographic factors, the comparison of data on patients with respect to certain types of NICUs and differences in the case-mix of obstetric and neonatal patients may influence the “center-effect.” In a study of 743 VON centers, data on Black, Hispanic, and Asian infants were evaluated across NICUs and large regional differences were observed in NICU quality.¹⁷ After accounting for these geographic differences, compared with White infants, Asian infants received care at higher-quality NICUs and black infants, at lower-quality NICUs. These factors highlight the relationships between quality of care, location of care and healthcare disparities. Factors such as race and ethnicity may also be associated with specific aspects of care, such as breastfeeding.¹⁸ Therefore, factors such as location of a unit, the types of patients cared for in a particular unit, and care practices, should be considered alongside each other.

Risk appropriate care

Births in NICUs with higher levels of care (Level III) and/or larger volumes of neonatal patients have been associated with lower risk-adjusted neonatal mortality.^19,20 Emphasis on risk-appropriate care for infants <32 weeks’ gestation and birth weights <1500 grams in level III neonatal intensive care units have been recommended by professional societies in the US,²¹ although implementation of the policies varies by state.²² Risk-appropriate care is needed to ensure that infants, such as those born at <25 weeks’ gestation, are cared for in centers that can provide sustained life support, have access to a full range of subspecialists, respiratory support, and advanced imaging. Similar guidance recommendation documents for regionalization and levels of care are present in other developed countries.²³ However, even among Level III neonatal intensive care units, large variabilities in both outcomes and practices are evident, as discussed in detail in sections below. This review will focus on the importance of center, as it pertains to infants cared for in Level III or higher centers, in view of the findings of higher mortality in lower level centers, particularly for the most immature infants.²¹

The importance of the center of birth is exemplified in a recent update of a widely used model to predict in-hospital survival that was developed by the NICHD Neonatal Research Network (NRN). Rysavy et al. reported that, after the birth weight of an infant (which contributed 36% to the prediction of survival), the hospital of birth contributed as much as did gestational age (20% each) to the prediction of survival, and more than infant sex, receipt of antenatal corticosteroids, or multiple gestation.²⁴ This model highlights the importance of consideration of center, when providing prognostic estimates for survival at 22 to 25 weeks’ gestation.

Approaches to prognosis, perinatal counseling, and shared decision-making

Prognosis can be considered an intervention,²⁵ in that such predictions may influence decisions to pursue or forgo resuscitative measures. Approaches to prognosis and perinatal counseling vary among centers, with few centers providing specialized training to designated prenatal counselors.²⁶ In addition, underestimation of favorable outcomes is associated with restriction in the use of resuscitative interventions among both neonatal²⁷ and obstetric providers.²⁸ Discussion and rates of withdrawal of life-sustaining treatment vary widely across centers and are associated with the patient’s race or ethnicity, with life-sustaining treatment withheld or withdrawn more commonly for white non-Hispanic infants, compared to non-Hispanic black and Hispanic infants at US centers.^29,30 While frameworks exist for shared decision-making in neonatal ICUs,³¹ how these approaches vary among units and explain differences in outcomes among centers remains poorly understood.

Approaches to Measurement and Reporting

Some variations may be due to differences in population coverage, transfer policies, delivery room deaths, outcome definitions, and the amount of missing data.³² Consistent reporting of outcomes of extremely preterm births may reduce biases in estimates, and potentially allow for more direct comparisons, to allow for consideration of differences in outcomes among centers.³³ While these potential sources of bias may influence the assessment and interpretation of center variation, variations in clinical practice also have important impacts on variations in clinical outcomes, as discussed in the next sections.

Obstetric Care Practices

In the following section, we focus on specific obstetric practices that are associated with survival or other important neonatal outcomes and vary in use among centers.

Antenatal Corticosteroid Use

Among obstetric care practices for the mother and fetus, provision of antenatal corticosteroids has been highlighted as one of the most important interventions to improve survival for infants born at <25 weeks’ gestation. Despite the known benefits of antenatal steroids on neonatal survival,³⁴ its use is variable, as highlighted by two studies from California,^35,36 with variations in use associated with geographic region and level of care.

More recently, a study of 431 US VON member hospitals demonstrated that receipt of antenatal steroids was associated with increased survival at each gestational week from 22 to 25 weeks among those infants that received postnatal life support.³⁷ For infants born at 22 weeks’ gestation, antenatal steroid use was associated with a greater than 2 fold increase in survival from 17.7% to 38.5% (adjusted relative risk 2.11; 95% CI 1.68–2.65). The association with antenatal steroid use and survival at 25 weeks’ gestation was still significant, but diminished in magnitude (relative risk of survival of 1.11; 95% CI 1.07–1.14). Overall, in this cohort, 52.4% of infants who received postnatal life support at 22 weeks’ gestation received antenatal steroids. By contrast, in the reports of centers from Iowa, Sweden and Uppsala, provision of antenatal corticosteroids was substantially higher, ranging from 85–93% (Table 2).

Cesarean Delivery

Although cesarean delivery may benefit survival for some infants at very early gestational ages, these benefits need to be considered alongside the risks to the mother.³⁸ In addition, data from randomized trials comparing cesarean to vaginal delivery for preterm birth are limited by small sample sizes, with imprecision in estimates and inclusion of a broader gestational range of preterm infants.³⁹ While cesarean delivery, compared to vaginal delivery, is not significantly associated with a reduction in perinatal mortality (relative risk of 0.29, 95% CI 0.07 to 1.14, based on three trials with 89 women), point estimates suggest the potential for benefit. Similar limitations are present in data to recommend immediate vs. deferred delivery, in a compromised preterm fetus.⁴⁰

In the US, joint guidance from the American College of Obstetricians and Gynecologists (ACOG) and the Society for Maternal-Fetal Medicine (SMFM) does not recommend cesarean delivery for fetal indication at 22 weeks’ gestation and notes consideration of cesarean delivery at 23 to 24 weeks, with a recommendation at 25 weeks.⁴¹ As mentioned previously, even among centers with high survival for infants born at 22 weeks’ gestation, rates of cesarean delivery range from 0 to 75% (Table 2).

Fetal Heart Rate Monitoring

Because electronic fetal heart rate monitoring is linked to plans for cesarean delivery due to fetal indications, this care practice is likely to be based on whether a cesarean delivery will be offered. However, the decision to provide fetal heart rate monitoring may reflect intentions for active obstetric treatment. The data are limited on whether fetal heart rate monitoring can provide important guidance regarding postnatal resuscitation of infants or can improve outcomes for infants born at <25 weeks’ gestation.

Other Obstetric Practices

Practices such as tocolysis, magnesium sulfate for neuroprotection, antibiotics to prolong latency, and cervical cerclage, may influence the timing of preterm birth or fetal or neonatal survival or morbidity, although recommendations for these practices vary by gestational age in the US, with most practices not recommended at 22 weeks’ gestation.⁴² In addition, progesterone may reduce the risk of preterm birth and improve perinatal outcomes,⁴³ although there is uncertainty in its potential effectiveness.⁴⁴ Variation in the use of these practices among centers may influence outcomes of births occurring at <25 weeks’ gestation.

Receipt of Prenatal Care

Prenatal care is important to ensure the ability to receive timely and necessary obstetric care and interventions, including those described above. Disparities in prenatal care may be linked to racial, ethnic, and socioeconomic disparities in obstetric care,⁴⁵ with younger maternal age and less than high school education being important factors associated with lack of prenatal care.⁴⁶ Removal of barriers in accessing prenatal care may depend on factors at the hospital, community, regional and national level, such as public health insurance programs.

Infant Care Practices

Infant care practices that might explain center variation in outcomes include both delivery room and neonatal intensive care unit practices (Figure 1). In addition, post-discharge follow-up care and interventions may potentially influence longer-term outcomes, but are beyond the scope of this review.

Active Treatment (Resuscitation)

One of the most important factors that explains center differences in outcomes at 22–23 weeks’ gestation is the provision of active treatment (resuscitation). In a study of 24 US hospitals that are part of the NICHD NRN, Rysavy et al. demonstrated marked variation in provision of active treatment, particularly at 22 and 23 weeks’ gestation, with interquartile ranges of active treatment from 7.7% to 100% at 22 weeks’ and 52.5% to 96.5% at 23 weeks’ gestation.⁴⁷ While most hospitals provided active treatment to all infants born at 25 or 26 weeks of gestation, only 5 of 24 hospitals provided active treatment to all infants born at 22 through 26 weeks’ gestation. The hospital rates of active treatment accounted for 78% of the between-hospital variation in survival among infants born at 22- or 23-weeks’ gestation.

These data indicate the provision of active treatment at a center is a large component of the “center effect.” Guidance from ACOG and SMFM in the US recommend consideration of active treatment at 22 and 23 weeks’ gestation, with a full recommendation beginning at 24 weeks’ gestation.⁴¹ Guidelines for the management of extremely premature deliveries in other very highly developed countries (based on rating by the United Nations Development Programme’s Human Development Index) vary, with most supporting comfort care at 22 weeks’ gestation.⁴⁸ Such guidelines might incorporate organizational, cultural, or social factors within countries, and reflect not only the importance of center of birth but also country or region of birth. Importantly, among countries, regional variation exists in both practices and outcomes, as highlighted by a study of differences in perinatal death among 7 different regions of Sweden.⁴⁹

In a study of US hospitals that were part of the VON, postnatal resuscitation occurred among 89% of all infants, but in less than 40% for infants at 22 weeks’ gestation.³⁷ Survival rates were 29% of infants at 22 weeks’ and 52% at 23 weeks’ gestation. By contrast, active treatment among 3 centers in Table 1 with >50% survival at 22 weeks’ gestation had active treatment rates of 91% (Iowa), 62% (Cologne) and 100% (Uppsala).

Linked to neonatal assessment for active treatment is the availability of a neonatologist, or similarly trained staff, in the delivery room. Appropriate staff in the delivery room may be a target for improving delivery room resuscitation.⁵⁰ In a UK study, consultant attendance at very preterm births varied markedly, and was more common in tertiary units.⁵¹ In a comparison of two centers (Uppsala in Sweden and Nationwide Children’s in the US), one of several differences between the two centers was neonatology presence in the delivery room, which was 100% at the center in Uppsala and 75% in the center at Nationwide (Table 1).¹⁴ However, the data are not sufficient to demonstrate that neonatologist presence in the delivery room is associated with improved survival at <25 weeks’ gestation.

Temperature Management

Hypothermia, specifically a temperature of <36.0° C upon admission to the neonatal unit, is associated with worse outcomes, including a higher risk of mortality and sepsis, with marked variation in admission temperature profiles among a study of 15 centers in the NICHD NRN.⁵² Improvements in admission hypothermia have been observed over time, although the association between lower admission temperatures and higher mortality has continued to be reported.⁵³ Although it is unclear if this association is causal, efforts to prevent heat loss in the delivery room are now part of routine resuscitation practice. Prevention of hypothermia was one of 4 key care domains highlighted in the EPICE study⁹ and is included in the Baby-MONITOR composite measure.¹⁰

Delayed Cord Clamping

Delayed cord clamping for infants born at <34 weeks’ gestation is recommended by the International Liaison Committee on Resuscitation, while cord milking is not recommended for infants born at <28 weeks,⁵⁴ in view of the higher risk of severe intraventricular hemorrhage observed in a randomized trial among infants born at 23 to 27 weeks’ gestation.⁵⁵ A systematic review and meta-analysis of 18 trials found delayed cord clamping reduces in-hospital mortality (relative risk [RR] 0.70; 95% CI 0.51–0.95).⁵⁶ The report by Mehler et al. highlights delayed cord clamping as one key feature of active prenatal and postnatal care at the University of Cologne.¹³

Approaches to Respiratory Support

Bronchopulmonary dysplasia (BPD) is a common morbidity among infants born at <25 weeks’ gestation, and a number of potentially better practices have been attempted to prevent BPD,⁵⁷ with limited success, as seen with the persistence of BPD over time among extremely preterm infants.⁵⁸ The European Consensus Guidelines on the Management of Respiratory Distress Syndrome⁵⁹ and the American Academy of Pediatrics Committee on Fetus and Newborn⁶⁰ recommend non-invasive approaches to respiratory support for spontaneously breathing infants, with selective surfactant administration. At Iowa and Uppsala, 100% of infants born at 22 weeks’ gestation received surfactant (Table 1), while in Cologne 75% received less invasive surfactant. Importantly, omission of surfactant treatment among ventilated infants is associated with lower survival, based on data from a Swedish cohort study.⁶¹ Therefore, optimal surfactant use among infants <25 weeks’ gestation is unclear, and emerging data suggest that less invasive surfactant administration may be associated with a lower risk of BPD or death,⁶² a key practice highlighted by Mehler et al., in their outcome report from Cologne, Germany.¹³

Use of Medications

A multicenter cohort of 7578 infants born at 22–24 weeks’ gestation demonstrated common exposure to multiple medications (median of 13 with interquartile range of 8–18), with variation in use among sites.⁶³ Ampicillin, Gentamicin, Surfactant, Dopamine and Caffeine were the five most commonly used medications at 22 and 23 weeks’ gestation, while Vitamin A use was uncommon.

Prolonged empiric antibiotic use, including ampicillin and gentamicin, has been associated with increased risks of necrotizing enterocolitis and death, with wide variation in empiric therapy among centers, ranging from 27–85%.⁶⁴ Although more recent studies have not shown an association with adverse outcomes including death or neurodevelopmental impairment,⁶⁵ nor death or necrotizing enterocolitis,⁶⁶ given the high antibiotic exposure and variation in use, this remains an important target for improvement efforts among centers with variation in adherence to best practices.⁶⁷ Dopamine and other inotropes used for antihypotensive therapy have been associated with an increased risk of death or neurodevelopmental impairment,⁶⁸ with marked center variation in use ranging from ~20% to 80% of infants born at 23 to 26 weeks’ gestation receiving antihypotensive therapy at NRN centers.⁶⁹

Caffeine has been shown to improve long-term neurodevelopmental outcomes.⁷⁰ Earlier initiation of caffeine, consistent with prophylactic use, has become more common over time.⁷¹ In a study of 62,056 very low birth weight infants from US NICUs, initiation of caffeine before 3 days of life was associated with a lower incidence of BPD, which was noted among subgroups of patients <24 weeks’ and 24–28 weeks’ gestation.⁷¹ By contrast, among infants delivered at <24 weeks’ gestation, an association between early caffeine use and increased mortality (odds ratio [OR] 2.76; 99% CI 1.77–4.28) was reported, which was not observed among infants at 24– 28 weeks’ gestation (OR 0.99; 95 CI 0.87–1.13). These findings highlight the importance of consideration of potential heterogeneity in effects of therapies amongst the most immature infants, including not only if they are used, but how they are used (e.g., timing of onset) in considering how variability in medication use among centers may influence outcomes.

Another therapy with variable use is Vitamin A, with its use associated with the belief of its efficacy by NICU medical directors.⁷² While Vitamin A is one of few medications shown to reduce the risks of BPD in clinical trials, a recent study highlights the potential heterogeneity in effect of Vitamin A therapy for the prevention of BPD, which showed a greater magnitude of treatment effect among infants with a lower risk of death or BPD than in those infants with higher risks of death or BPD.⁷³ This would suggest that the most immature infants, with the highest risk of death or BPD, are the least likely to benefit from Vitamin A therapy.

Finally, use of probiotics has been increasing among US centers,⁷⁴ and data from meta-analysis of clinical trials⁷⁵ and cohort studies of routine use⁷⁶ both indicate favorable benefit in reducing necrotizing enterocolitis and, potentially, death. However, some uncertainty persists in the efficacy and safety among the most immature infants, including those <25 weeks’ gestation. How these medications are used among centers may reflect on the incidence of associated outcomes, although these are only some of the factors that may influence the “center effect.” It is also important to consider not only variability of use between centers, but also within a center based on a clinician’s perceptions of benefits or harms for use in infants <25 weeks’ gestation, which may differ from the overall use of medications at a given center.

Human Milk Feeding

Human milk feeding is the optimal strategy to reduce the risk of necrotizing enterocolitis,⁷⁷ with a higher dose of human milk intake in the first 14 days of life associated with lower risk of death or necrotizing enterocolitis.⁷⁸ Human milk feeding at discharge is included in the Baby-MONITOR composite measure¹⁰ and was considered for the EPICE cohort study⁹ but not included to limitations in measurement.

Conclusion

The “center-effect” may reflect variation in obstetric and neonatal care practices that have been highlighted in Figure 1, as well as a multitude of other factors, including national and regional characteristics, risk-appropriate care, measurement of outcomes, quality, case-mix within a center, and perceptions of prognosis and approaches to shared decision-making. These factors, considered together, may explain why large variations are observed in both clinical practices and outcomes for extremely preterm infants, including those born at <25 weeks’ gestation. Characteristics associated with centers that have high survival rates of births at <25 weeks gestation, particularly those centers with high survival rates for infants born at 22 and 23 weeks’ gestation, may guide development of care practices to improve outcomes and inform hypotheses for future studies. However, these practices should not be considered in isolation and need to be viewed in the full context of the available evidence and the local context of care, given the inter-related nature of the factors. Future research is necessary to better understand how the structure and process of care in centers, along with the types of patients and families cared for, influence neonatal outcomes.