Predicting Long-Term Neurodevelopmental Outcomes in Very Preterm Neonates by Umbilical Cord Gas Parameters

Abstract

Background:

The predictive value of acidemia at birth for long-term neurodevelopmental outcomes remains poorly understood, especially in preterm neonates.

Objective:

To assess the relationship between umbilical artery acid-base status and major neurodevelopmental disability at age 5–8 years among children born very preterm.

Study Design:

We performed a secondary analysis of a follow-up study of a prospective cohort of 457 children between 23 weeks and 31 weeks 6 days from 1996–2001. Arterial cord gas parameters <10^th percentile for the original cohort of 457 neonates: pH < 7.1, base deficit <−8.6 mEq/L, and PCO2 > 77mmHg were considered abnormal. Sensitivity analyses considered alternative definitions of abnormal cord gases including pH < 7.0 or base deficit <−12. The primary outcome was a composite of major neurodevelopmental disability, including IQ < 70, cerebral palsy, blindness, deafness, abnormal balance, impaired cognition, dystonia, and seizure disorder. Logistic regression was used to adjust for race and caregiver IQ and, in an additional analysis, for gestational age.

Results:

259 of 261 maternal-infant dyads were evaluated at a mean child age of 6.8 years, of which 228 had complete cord gas data. Infants with abnormal pH and base deficit (defined above) were over four times more likely to have the composite disability, as well as IQ < 70. These increased odds persisted after adjustments for age and caregiver IQ, but when considering gestational age as well, none of the cord gas parameters significantly predicted presence of the composite disability or IQ <70. However, when using stricter cord gas criteria (pH < 7.0, base deficit <−12), base deficit < −12, was independently associated with both neurodevelopmental disability and IQ < 70.

Conclusion:

When defined more strictly, abnormal cord gases, specifically base deficit <−12, are associated with an increased risk of long-term major neurodevelopmental disability and IQ<70 in children born very preterm.

Introduction:

It has been suggested that umbilical cord gases be obtained after the birth of any high-risk or depressed neonate to assist in clinical assessment of the newborn.¹ Prior studies have evaluated, which, if any, parameter of the blood gas analysis serves as the best predictor of short-term morbidity.^2–6 However, the predictive value of acidemia at birth for long-term childhood neurodevelopmental outcomes remains poorly understood. Although severe acidosis at birth in cases of suspected hypoxic-ischemic encephalopathy has been reported as a strong predictor of death and neurodevelopmental impairment in term infants (not older children),⁷ this is not universally observed. In fact, most acidemic neonates are neurologically normal.⁸ Furthermore, even less is understood about the predictive value of severe acidosis in preterm infants, a group in which a multitude of factors, most notably gestational age at delivery, is likely to influence long-term neurodevelopment.⁷

Several studies have suggested that acidemia at birth in infants born at <32 weeks is associated with a higher risk for short-term adverse outcomes but not long-term adverse neurodevelopmental outcomes.^9,10 Although a recent large study suggested a higher risk for death and neurodevelopmental impairment in extremely low birthweight (<1000 g) infants with perinatal acidosis, this study limited follow-up to 18–22 months corrected age.⁷ Therefore, the more long-term effects of perinatal acidosis remain largely unknown. As such, we aimed to use the previously described NICHD PERC follow up study cohort¹¹ to assess the relationship between umbilical artery acid-base status and major neurodevelopmental disability at age 5–8 years among children born very preterm (VPT).

Materials and Methods:

We performed a secondary analysis of the data previously collected by NICHD PERC follow up study¹¹ of the Alabama Preterm Birth Study^12–18. The original Alabama Preterm Birth Study was a prospective cohort enrolling 457 consecutive, singletons born between 23 and <32 weeks from 1996 to 2001 at the University of Alabama at Birmingham in order to evaluate risk factors and perinatal outcomes after preterm birth. Details of the study design, definitions of maternal and neonatal outcome variables, and results have previously been reported (UAB; HD33927).^12–18 In this original cohort study, trained research nurses collected detailed pregnancy and neonatal (birth to discharge or death) outcome data via chart review for each of the initial maternal-infant dyads. Detailed information regarding delivery indications, labor data, umbilical cord data, and neonatal complications was collected and analyzed to determine potential relationships between neonatal outcomes and other factors (i.e interleukin-6 concentrations, placenta infiltrates, placental histology, and umbilical cord cultures).^12–18 The NICHD PERC follow up study enrolled 261 maternal-infant dyads of the original 457 in order to assess long-term outcomes in the offspring born from the Alabama Preterm Birth Study.¹¹ In this follow-up study, original participants were contacted by research personnel and consented to participate in a series of tests in order to assess for the presence of severe neurodevelopmental disability in the children who were 5 to 8 years old at that time. Primary caregivers, generally the child’s mother, underwent a variety of psychometric measures and the Peabody Vocabulary Test 3^rd edition (PPVT-III)¹⁹, which served as a proxy for caregiver IQ. The children also underwent a variety of psychometric measures, including the Wechsler Intelligence Scale for Children-IV²⁰ or Differential Ability Scales²¹ depending on the child’s age, to evaluate IQ. Each child was also subject to a physical and neurological exam. Infants with known neurodevelopmental disorders/anomalies at birth were excluded.

All 261 maternal-infant dyads enrolled in the NICHD PERC follow up study were considered for possible inclusion in this study. As the data was limited to the follow up data within the NICHD PERC follow up study, the study population was defined as above as a sample of convenience. Each maternal-infant dyad from the follow up study was successfully linked to arterial cord blood gas, but diagnosis of the composite outcome was missing from two dyads, resulting in their exclusion.

The remaining 259 dyads were classified by multiple umbilical artery cord gas parameters, including pH (pHa), base deficit, and PCO2 based on whether or not each of these parameters was normal or abnormal. Arterial cord gas parameters <10^th percentile for the original cohort of 457 neonates, defined as pH <7.1, base deficit <−8.6 mEq/L, and PCO2 >77mm Hg were considered abnormal. Given updated guidelines from the American College of Obstetricians and Gynecologists (ACOG), a secondary analysis defined abnormal cord gases as pH < 7.0 or base deficit <−12.²² However, even data in the updated guidelines from ACOG shows the presence of morbidity at the less strict pH cut off of 7.1 which prompted our use of less strict cord gas parameters.⁸ The primary outcome was a composite of major neurodevelopmental disability (NDD) at age 5–8 years, as previously defined by Andrews et al.¹¹ This composite included IQ < 70, cerebral palsy, blindness, deafness, or other major deficits such as abnormal balance, impaired cognition, dystonia, and seizure disorder. Children were considered to have the composite outcome if they had one or more of the above inclusion diagnoses or deficits. IQ was derived from the child’s score on the WISC-IV or DAS. Cerebral palsy was defined as abnormal muscle tone in at least one extremity and abnormal control of movement or posture.¹¹ IQ < 70 was the only secondary outcome of interest, thus, an additional analysis assessing the relationship between arterial cord gas parameters and IQ alone was also performed.

For the primary and additional analyses, children with the primary and secondary outcomes were compared using chi-square or Fisher exact tests as appropriate. A p-value of <0.05 was considered statistically significant. Odds ratios (OR) with 95% confidence intervals (CI) were calculated. In addition, logistic regression was used to adjust for race (known association with neurodevelopment and IQ)²³ and caregiver IQ (Model 1); an additional analysis also adjusted for gestational age, race, and caregiver IQ (Model 2). As birthweight is highly correlated with gestational age, it was not included to account for over-adjustment. Maternal morbidities (hypertension, diabetes) were also not considered covariates as there is little biologic plausibility for maternal medical conditions to affect childhood neurodevelopment independent of gestational age at delivery. Similarly, APGAR scores were not considered covariates as they have not been shown to predict neurologic outcome in neonates.²⁴ SAS software version 9.3 (Cary, NC) was used to complete all analyses in this study. No adjustments were made for multiple comparisons.

Results:

Of the 259 maternal-infant dyads considered for inclusion in this study, 41 dyads were noted to have composite neurodevelopmental disability (NDD), while the remaining 218 did not. Baseline demographic and other characteristics by presence or absence of the primary outcome (NDD vs. No NDD) are presented in Table 1. Those children with the primary outcome were of earlier gestations (26.9 weeks vs 28.6 weeks, p<0.0001) and lower birth weight (912.4g vs 1211.1g, p<0.0001). They were also significantly more likely to have lower 1 and 5 minute APGAR scores. However, there was no significant difference in maternal age, BMI, marital status, child’s reported race, or in child sex. While there was not significant difference in maternal education level, there was a statistically significant difference in primary caregiver IQ (87.5 vs 92.2, p 0.02). As a result, primary caregiver IQ and gestational age were considered covariates. As detailed above, race was also considered a covariate, but APGAR scores and birthweight were not.

Table 1.

Demographics of patients and covariates by presence or absence of composite neurodevelopmental disability (NDD)

	NDD (n=41)	No NDD (n=218)	p-value
Maternal Years of Age, mean ± sd	24.2 ± 5.2	25.1 ± 6.4	0.37
Child Race			0.93
Black	30 (73%)	128 (59%)
Non-Black	11 (27%)	90 (41%)
Education			0.82
<12 years	10 (28%)	56 (30%)
12+ years	26 (72%)	133 (70%)
Tobacco Use in Pregnancy	2 (5%)	20 (9%)	0.54
Married at Delivery	11 (28%)	93 (43%)	0.07
Maternal BMI (kg/m2), mean ± sd††	31.0 ± 8.7	31.1 ± 9.0	0.99
Primary caregiver IQ	87.5 ± 11.5	92.2 ± 11.2	0.02
Preeclampsia	15 (37%)	71 (33%)	0.62
Clinical chorioamnionitis	5 (12%)	26 (12%)	>0.99
PTB classification			0.02
Spontaneous	19 (46%)	61 (30%)
Indicated	17 (42%)	76 (36%)
PPROM	5 (12%)	70 (33%)
Parity			0.69
Nulliparous	19 (46%)	108 (50%)
Multiparous	22 (54%)	109 (50%)
GA at delivery (weeks), mean ± sd	26.9 ± 2.2	28.6 ± 2.1	<0.0001
Female Child Sex	21 (51%)	122 (56%)	0.58
Birthweight (grams), mean ± sd	912.4 ± 331.1	1211.1 ± 373.7	<0.0001
1-minute APGAR, median (Q1, Q3)	5 (2, 6)	6 (4, 7)	0.005
5-minute APGAR, median (Q1, Q3)	7 (6, 8)	8 (7, 9)	0.01
Antenatal corticosteroids	36 (88%)	189 (88%)	0.96
Antenatal magnesium sulfate	14 (35%)	70 (33%)	0.82
SGA (BW<10th percentile)	5 (12%)	16 (7%)	0.35

^*Table includes all 259 women from the primary PERC study with neurodevelopmental outcomes. Data does not exclude individual patients with missing cord gas information considered for analysis in this study.

Chorioamnionitis and IUGR were balanced between groups and are the focus of an already published study¹¹

Of the 259 maternal-infant dyads, 228 had complete cord gas data and were further analyzed for relationships with NDD. Incidences of both the composite disability (NDD) and IQ<70 by abnormal umbilical cord gas parameters are presented in Table 2. Children with abnormal pH, base deficit, and PCO2 all had significantly higher rates of NDD than those children with normal cord gas parameters. Children with either pH<7.1 or base deficit <−8.6mEq/L were over four times more likely to be diagnosed with the composite disability than their counterparts with normal cord gases (unadjusted OR 4.3 [95% CI 1.5–12.5[; OR 4.4 [1.5 – 12.6] respectively), while children with PCO2 >77 had over a two-fold risk of composite disability (OR 2.6 [1.03–6.5]). Similar results were demonstrated between these abnormal blood gas parameters and the secondary outcome of IQ<70 (Table 2).

Table 2.

Incidence of composite neurodevelopmental disability (NDD) and IQ <70 by umbilical cord gas parameter.

Outcome: NDD
Parameter/Risk factor	Rate with risk factor	Rate without risk factor	p-value	Unadjusted OR
pH <7.1	7 (44%)	32 (15%)	<0.01	4.3 (1.5 – 12.5)
Base deficit <-8.6 mEq/L	7 (44%)	30 (15%)	<0.01	4.4 (1.5 – 12.6)
PCO2 >77mmHg	8 (32%)	31 (15%)	0.04	2.6 (1.03 – 6.5)
Outcome IQ<70
Parameter/Risk factor	Rate with risk factor	Rate without risk factor	p-value	Unadjusted OR
pH <7.1	6 (38%)	24 (11%)	0.01	4.7 (1.6–14.0)
Base deficit <−8.6 mEq/L	6 (38%)	22 (11%)	<0.01	4.7 (1.6 – 14.3)
PCO2 >77mmHg	7 (28%)	23 (11%)	0.03	3.0 (1.1 – 8.0)

The associations between long-term childhood neurodevelopmental disability and umbilical artery parameters at birth when adjusted for significant covariates are presented in Table 3. When adjusting for race and maternal IQ (Model 1), there were still significantly increased odds of composite NDD associated with abnormal pH (aOR 3.6 [1.2–11]) and base deficit (aOR 3.7 [1.2–11]). Similar findings were seen in Model 1 for IQ < 70. However, there was no significant associations of NDD or IQ< 70 with PCO2 >77. Yet when additionally adjusting for gestational age as well as race and maternal IQ (Model 2), none of the umbilical artery cord gas parameters were significantly associated with composite NDD or IQ<70. In fact, when evaluating the odds of composite NDD based on gestational age alone, increasing gestational age (by week) was significantly associated with a 25–30% decrease in the odds of composite NDD even when umbilical cord gas parameters were abnormal (aOR range 0.7–0.75). Similar results were demonstrated when considering the secondary outcome of IQ <70 alone (Results not shown for GA only model for IQ). Of note, there were no significant interactions between each cord gas value and birth GA (all p>0.05) to justify any further stratified analysis based on early (23–28 weeks) versus late (>28 weeks) birth GA.

Table 3.

Adjusted odds of long-term NDD and IQ < 70 based on umbilical artery narameters at birth (Data nresented as adjusted odds ratios with 95% confidence intervals)

Parameter	pH <7.1	Base deficit < −8.6 mEq/L	PCO2 > 77 mm Hg
NDD
Model 1*	3.6 (1.2- 11)	3.7 (1.2-11)	2.2 (0.8–5.8)
Model 2**	2.4 (0.7– 7.6)	2.5 (0.8– 8.2)	1.9 (0.7– 5.3)
IQ < 70 only
Model 1*	4.1 (1.2- 14)	4.4 (1.3- 15)	2.6 (0.9–7.3)
Model 2**	2.6 (0.7– 9.1)	2.8 (0.8– 12)	2.2 (0.7– 6.5)

Unadjusted odds ratios from Table 2 are:

^*Adjusted for race and maternal IQ.

^**Adjusted for race, maternal IQ, and gestational age.

Additional analysis based on alternative abnormal cord gas parameter definitions (pH < 7.0 or base deficit <−12) is presented in Table 4. There were significantly increased odds of both composite NDD and IQ<70 when evaluating pH < 7.0, base deficit < −12, and pH < 7.0 or base deficit < −12. When adjusting for race and maternal IQ (Model 1), there were still significantly increased odds of composite NDD and IQ < 70 associated with these three cord gas parameters (Table 4). However, when additionally adjusting for gestational age, only base deficit <−12 was significantly associated with both composite NDD (aOR 7.8 [1.3–46.0]) or IQ<70 (aOR 6.0 [1.1–33.5]). When looking at either pH <7.0 or base deficit <−12, similar findings were noted for NDD after adjusting for race, maternal IQ, and gestational age, but not for IQ <70.

Table 4.

Odds of long-term NDD and IQ < 70 based on pH < 7.0 and Base deficit (BD) < -12 (Alternate Analysis; data presented as unadjusted or adjusted odds ratios with 95% confidence intervals)

Parameter	pH <7.0	Base deficit < -12 mEq/L	pH < 7.0 or BD < -12
NDD
Unadjusted model	7.0 (1.5-32.7)	17.0 (3.3-87.8)	11.6 (2.8-48.9)
Model 1*	6.4 (1.4-30.6)	13.5 (2.4-74.9)	9.1 (2.0-41.3)
Model 2**	4.5 (0.9–23.0)	7.8 (1.3-46.0)	5.9 (1.2-28.4)
IQ < 70 only
Unadjusted model	5.3 (1.1-25.1)	13.1 (2.9-58.4)	10.2 (2.6-40.8)
Model 1*	5.5 (1.1-27.5)	10.7 (2.1-55.2)	8.6 (1.9-39.8)
Model 2**	3.7 (0.7–19.8)	6.0 (1.1-33.5)	5.3 (1.0–27.1)

Unadjusted model describes unadjusted odds ratios of outcome by alternate definitions of cord gas parameter.

^*Unadjusted model OR adjusted for race and maternal IQ.

^**Unadjusted model OR adjusted for race, maternal IQ, and gestational age.

Structured Discussion/Comment:

Principal Findings

In this cohort, abnormal (decreased) pH and base deficit below the 10^th percentile were not significantly associated with development of NDD or IQ<70 after adjusting for race and maternal IQ, and gestational age. However, when considering stricter abnormal cord gas parameters as supported by ACOG, base deficit <−12 (and base deficit <−12 or pH < 7.0) were independently associated with NDD and IQ < 70. This suggests that perhaps the more extreme cord gas parameters, rather than those marking the 10^th percentile, should be those noted to assess for at risk children born very preterm.

Results

While the predictive value of cord gas parameters on short-term disability has been evaluated by previous studies, the effect on long-term childhood neurodevelopmental outcomes has been less studied, especially in children born very preterm. One study using the NICHD NRN preterm infant registry assessed the association between perinatal acidosis and risk for death and neurodevelopmental impairment in extremely low birth weight infants⁷. This study showed a significant relationship between perinatal acidosis (defined as pH <7.0 or base deficit <−12) and death and neurodevelopmental outcomes in extremely low birth weight neonates, but noted that other unspecified factors may be more important in predicting these outcomes. While this study was much larger, including nearly 4000 infants, study follow up was limited to only 22–26 months chronological age (infancy). Further, the diagnosis of neurodevelopmental impairment within that study differed based upon the year of delivery.⁷ While our study included fewer participants, follow up was obtained at 5–8 years, allowing for evaluation of more long-term neurodevelopmental outcomes extending into childhood, rather than infancy. Reasons for need to assess neurodevelopment in childhood, versus just infancy, include impact of education, different neurologic assessments targeted at school-age children obviating the need for parent responses, and the possibility that potential neurodevelopmental deficits apparent in infancy may improve with identification and early intervention (physical, speech, occupational therapy). These long-term outcomes are perhaps more impactful to parents than shorter term outcomes that can be actionable.

Clinical Implications

As such, we show that base deficit <−12 alone and base excess <−12 or pH<7.0 at birth in very preterm infants are associated with long-term childhood neurodevelopmental disability. Thus, there is potential need for additional neurodevelopmental support for these at-risk infants.

Strengths and Limitations

There are several strengths of our study. The first strength is the long term follow up at a mean age of 6.8 years (school age range) for each of the infants included in the NICHD PERC follow up study. This is a well-characterized cohort that was followed prospectively from the initial 457 singleton birth study to the follow up study with detailed chart abstraction and follow up information. The follow-up included standard childhood testing procedures conducted by trained neurodevelopmental staff and reviewed by a “gold standard” examiner for proficiency. Nurses performing data collection were trained to ensure standardized data collection. Over 95% of those eligible for the study had data available, minimizing drop out bias. Finally, caregiver IQ was included in the multivariable models in an attempt to control for caregiver effects on the outcomes, especially in the setting of statistically significant differences in caregiver IQ between the two groups.

The limitations of this study include that it is a secondary analysis; this limited the ability to include data outside of that previously collected at the time of the initial study. For example, resuscitative maneuvers were not collected at the time of the initial study. However, all resuscitative efforts in this study were done at a single center with a single neonatology division with a well-guided protocol for resuscitation, minimizing deviations between providers, and thus, the efforts received by neonates in this study. In addition, while we adjusted for race, caregiver IQ, and gestational age in the logistic regression, there may be other unmeasured covariates (or covariates with low occurrences in our dataset i.e. intraventricular hemorrhage and periventricular leukomalacia) that contributed to the long-term outcomes. Furthermore, as with all long-term follow up studies, the inability to account for all intervening exposures is a limitation of this study. Finally, the overall sample size of 228 is relatively small, largely because severe neonatal acidemia is fortunately uncommon.

Research Implications

Future directions would include conducting a validation study in a cohort assembled from multiple sites. However, given the relatively large sample size of infants at 5–8 years of age, reproducing such a study would be exceedingly difficult and costly.

Conclusions

In summary, this study suggests that for children born VPT, using the ACOG definition of abnormal pH (<7.0) or base deficit (<−12) is independently predictive of an increased risk of long-term neurodevelopmental disability in children aged 5–8 years. Base deficit <−12 also increases the risk of IQ<70. Using cord gas abnormalities in the 10^th percentile, however, does not predict the risk of long-term neurodevelopmental disability after adjusting for gestational age. Therefore, this study suggests that the more extreme ACOG definitions for abnormal cord pH and base deficit may be more useful in predicting an increased risk for longer term neurodevelopmental disability in children born very preterm and allow for targeting of early intervention to reduce disability rates in these children.

Condensation.

Abnormal cords gases, specifically base deficit <−12, are associated with long-term major neurodevelopmental disability and IQ<70 in children born very preterm.

AJOG at a Glance.

1. Why was this study conducted?

   • To assess the relationship between umbilical artery acid-base status and major neurodevelopmental disability at age 5–8 years among children born very preterm
2. What are the key findings?

   • Abnormal cords gases, specifically base deficit <−12, are associated with long-term major neurodevelopmental disability and IQ<70 in children born very preterm. Less strict cord gas parameters were not predictive of neurodevelopmental disability or IQ<70.
3. What does this study add to what is already known?

   • This study suggests that the more extreme ACOG definitions for abnormal cord pH and base deficit may be more useful in predicting an increased risk for longer term neurodevelopmental disability in children born very preterm and allow for targeting of early intervention (physical, occupational, and speech therapy) to reduce disability rates in these children.

Highlights.

• This study suggests that for children born VPT, using the ACOG definition of abnormal pH (<7.0) or base deficit (<−12) is independently predictive of an increased risk of long-term neurodevelopmental disability in children aged 5–8 years.

• Using cord gas abnormalities in the 10^th percentile, however, does not predict the risk of long-term neurodevelopmental disability after adjusting for gestational age.

• The more extreme ACOG definitions for abnormal cord pH and base deficit may be more useful in predicting an increased risk for longer term neurodevelopmental disability in children born very preterm and allow for targeting of early intervention to reduce disability rates in these children.