Increased Risk of Adverse Neurological Development for Late Preterm Infants

Article Outline

• Abstract
• Methods
  • Study Population
  • Ascertainment of Neurological Development
  • Audit of Electronic Data
  • Data Analysis
• Results
  • Analysis Dataset
  • Maternal and Infant Characteristics
  • Distribution of Neurological Outcomes by Cohort Characteristics
  • Results of Manual Audit of Randomly Selected Records
  • Increasing Risk of Adverse Outcome with Decreasing Gestational Age
• Discussion
• Acknowledgment
• References
• Copyright

Objective

To assess the risks of moderate prematurity for cerebral palsy (CP), developmental delay/mental retardation (DD/MR), and seizure disorders in early childhood.

Study design

Retrospective cohort study using hospitalization and outpatient databases from the Northern California Kaiser Permanente Medical Care Program. Data covered 141 321 children ≥30 weeks born between Jan 1, 2000, and June 30, 2004, with follow-up through Jun 30, 2005. Presence of CP, DD/MR, and seizures was based on International Classification of Diseases, Ninth Revision codes identified in the encounter data. Separate Cox proportional hazard models were used for each of the outcomes, with crude and adjusted hazard ratios calculated for each gestational age group.

Results

Decreasing gestational age was associated with increased incidence of CP and DD/MR, even for those born at 34 to 36 weeks gestation. Children born late preterm were >3 times as likely (hazard ratio, 3.39; 95% CI, 2.54-4.52) as children born at term to be diagnosed with CP. A modest association with DD/MR was found for children born at 34 to 36 weeks (hazard ratio, 1.25; 95% CI, 1.01-1.54), but not for children in whom seizures were diagnosed.

Conclusions

Prematurity is associated with long-term neurodevelopmental consequences, with risks increasing as gestation decreases, even in infants born at 34 to 36 weeks.

Abbreviations: CP, Cerebral palsy, DD/MR, Developmental delay or mental retardation, ICD-9 CM, International Classification of Diseases, Ninth Revision, Clinical Modification, KPMCP, Kaiser Permanente Medical Care Program, LGA, Large for gestational age, SGA, Small for gestational age

Infants born prematurely (<37 completed weeks gestation) are at increased risk of death and disability.¹, ² In 2004, >500 000 infants, 12.5% of live births in the United States, were preterm,³ with the rate of such births increasing for >2 decades.² Births occurring between 34 and 36 weeks, sometimes labeled “near term,” but now preferentially referred to as “late preterm,”⁴, ⁵ comprise >70% of all preterm births and account for most of the increase in preterm birth rates.³, ⁶ Despite the growing prevalence of this group, research on morbidity and mortality in preterm births remains focused on the highest risk births, those occurring at <32 completed weeks gestation (very preterm).⁷, ⁸, ⁹ Much less is known about contemporary outcomes of late preterm infants, beyond infant mortality rates, which are intermediate between infants born very preterm and infants born at term (37-41 weeks).¹⁰, ¹¹, ¹²

The impact on child health from late preterm birth has only recently received broad attention,², ¹³, ¹⁴ including the definition, terminology, clinical care, knowledge gaps, and research priorities.¹³ Late preterm infants have higher rates of immediate newborn morbidities, such as respiratory distress syndrome, apnea, transient tachypnea of the newborn, hypoglycemia, hypothermia, prolonged jaundice, seizures, and feeding problems.¹⁵, ¹⁶ Late preterm infants are also more likely than term infants to require additional resource use, such as supplemental oxygen support, re-hospitalization, and higher medical care expenditures.¹⁷, ¹⁸, ¹⁹, ²⁰, ²¹, ²², ²³, ²⁴

Few studies have provided data related to neurodevelopmental outcomes in late preterm infants receiving contemporary neonatal care.², ¹³, ²⁵, ²⁶ To address this knowledge gap, we have analyzed inpatient and outpatient data for a cohort of infants born in the Kaiser Permanente Medical Care Program (KPMCP), an integrated healthcare delivery system with information systems built around a common medical record number for its 3.3 million members in Northern California. We assessed the association between moderate prematurity and the incidence of adverse neurodevelopmental outcomes, focusing on the diagnoses of cerebral palsy (CP), developmental delay/mental retardation (DD/MR), and the presence of seizure disorders.

Back to Article Outline

Methods 

Study Population 

The sample for this study was drawn from the Northern California KPMCP, the characteristics and information systems of which were described in detail in earlier reports.¹⁷, ²⁰, ²¹, ²², ²³, ²⁷, ²⁸, ²⁹ To be eligible for the study, children had to: 1) born alive at 1 of the 12 KPMCP birth facilities (the KPMCP medical centers at Fresno, Hayward, Oakland, Redwood City, Sacramento, San Francisco, Santa Clara, Santa Rosa, Santa Teresa, South Sacramento, Vallejo, and Walnut Creek) between Jan 1, 2000, and Jun 30, 2004; 2) survive the birth hospitalization; 3) have a gestational age at birth of at least 30 weeks; and 4) remain a member of the Kaiser Foundation Health Plan for at least 1 day after discharge from the birth hospitalization. All pregnancies to women in the KPMCP are dated by using ultrasound scanning between 12 and 24 weeks, and gestational length is expressed in the data system as completed weeks.

With methods we have described previously,¹⁷, ²⁰, ²¹, ²², ²³, ²⁷, ²⁸, ²⁹ eligible children were identified from the KPMCP hospitalization database. To obtain information on the conditions of interest, outpatient and inpatient utilization records were linked for the period between Jan 1, 2000, and Jun 30, 2005. Depending on the date of birth in this study period, follow-up could be as long as 5.5 years (Appendix I; available at www.jpeds.com). The integrated KPMCP information systems also included services provided outside KPMCP that were covered by the Kaiser Foundation Health Plan. Data for these analyses included the date of outpatient or inpatient events and the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9 CM) codes assigned at each encounter.

Appendix I. Neurological outcomes by follow-up time

	No CP, DD/MR, or seizures		CP		DD/MR		Seizures
	Number	Percent	Number	Percent	Number	Percent	Number	Percent
Follow-up time (years)
Up to 1	25380	18.2	19	5.3	1	0.1	6	3.4
1-2	38152	27.4	71	19.9	71	5.3	17	9.5
2-3	29939	21.5	81	22.7	261	19.5	46	25.7
3-4	22420	16.1	89	24.9	437	32.6	50	27.9
4-5	18420	13.2	76	21.3	447	33.4	41	22.9
5-5.5	5134	3.7	21	5.9	123	9.2	19	10.6
Total	139445	100.0	357	100.0	1340	100.0	179	100.0

The distribution of follow-up time for the CP, DD/MR, and seizure groups was statistically different (P < .001) from the group without any of these conditions identified.

This study was approved by the KPMCP Institutional Review Board for the Protection of Human Subjects, which has jurisdiction over all KPMCP facilities in Northern California.

Ascertainment of Neurological Development 

Neurological diagnoses were made on the basis of ICD-9 CM codes from patient encounter data. Because overlap between diagnoses could exist and the presence of 1 diagnosis (eg, CP) could be related to the presence of another (eg, seizures), the following outcomes hierarchy and sequential classification system was created to avoid double counting. First, infants with any of the ICD-9-CM codes 343.0 to 343.9 for at least 1 health encounter were identified and classified as having CP. Infants so identified were removed from the pool eligible for the next step in the algorithm. Second, of the remaining infants, those with ICD codes associated with DD/MR were identified when ≥2 coded encounters occurred at least 6 months apart; the DD/MR codes were those for specific delays in development (315.0-315.9) or mental retardation (MR; 317-319). Third, in the remaining children, those with ≥2 occurrences of ICD codes were identified, again separated by at least 6 months and associated with seizures, for epilepsy (345.0-345.9) or other convulsions (780.39). A 6-month interval was required between instances of ICD codes for DD/MR and seizures for several reasons: to eliminate those situations in which a physician assigned a diagnosis and then subsequently found that the first diagnosis had either resolved or only been provisional; to allow time for a subspecialty referral, when that was required; and to eliminate infants who might have had multiple visits for an initial diagnostic work-up. All remaining children (ie, those without any of the 3 identified conditions) were assigned to a fourth “no CP, DD/MR, seizures” group.

Additional demographic data were retrieved from the medical record: maternal race/ethnicity on the basis of the mother's self-report, was coded as Hispanic, black, Asian, white, or “other”; child's sex; plurality as singleton or multiple birth; small for gestational age (SGA) and large for gestational age (LGA), defined as birth weight <10th percentile and >90th percentile for gestational age, respectively; mode of delivery as either vaginal or cesarean; the 5-minute Apgar score. Dataset limitations precluded analysis of induction of labor and maternal ICD codes, including the indications for induction of labor and cesarean section.

Audit of Electronic Data 

To assess the robustness of the data, 2 researchers (G.J.E., M.L.M.) and an experienced neonatal nurse audited the data from 20 randomly selected medical record numbers for each of the 4 study groups (CP, DD/MR, seizures, or none of the 3 outcomes). The results of this audit are included in the sensitivity analysis (described in the Results section).

Data Analysis 

All analyses were conducted with SAS software version 9.1 (SAS Institute, Cary, North Carolina). For visual display of rates of neurodevelopmental outcomes at each gestational week, the moving average method was used, which both permits assessment of trends and compensates for infrequent occurrence of outcomes by smoothing fluctuations in a given rate.³⁰ This approach is also useful because of the universal uncertainty around gestational age dating. For example, the mean of the 3 crude rates for gestational ages 33, 34, and 35 weeks were used to obtain the CP rate moving average for infants of 34 weeks gestation. Moving averages were not used in the multivariate analyses, which instead used gestational age ranges (eg, 30-33, 34-36, 37-41, and 42+ weeks).

The Pearson χ² square value was calculated to compare the distribution between groups of the duration of clinical follow-up in the KPMCP system. To control for varying lengths of follow-up in the cohort, Cox proportional hazard models were used. In these models, the duration of follow-up from the first time an eligible diagnosis appeared in a given infant's encounter data is compared with the duration for infants without any of these conditions. Separate models for each condition (CP, DD/MR, and seizures) were generated, and associated crude hazard ratios for gestational age groups were calculated and then adjusted for relevant maternal and infant characteristics available in the database: maternal race/ethnicity, sex, plurality, and size for gestational age status.

The assumption of proportional hazards was tested by creating an interaction term for each variable with the log of the time variable. On the basis of the audit results (described in the Results section), the strength of our inference was further tested by assessing the sensitivity of the models. Some 25% of CP and DD/MR cases and 50% of seizures cases were randomly selected and reassigned to the group with no CP, DD/MR, or seizures. This was done 100 times for each model, and the resulting mean hazard ratios and confidence intervals were compared with the results of the original model (Appendix II; available at www.jpeds.com). The proportion of records set to “no delay” (ie, no CP, DD/MR, or seizures) in the CP, DD/MR, and seizure groups was determined by applying the respective percentage of records that could not be confirmed in the record audit.

Appendix II. Sensitivity testing of final models—resulting mean crude and adjusted hazard ratios^⁎

	CP		DD/MR		Seizures
	Hazard ratio	95% CI	Hazard ratio	95% CI	Hazard ratio	95% CI
Crude hazard ratios
Gestational age (weeks)
30-33	9.02	5.97-13.64	2.21	1.51-3.22	3.64	1.39-9.83
34-36	3.70	2.68-5.10	1.36	1.08-1.72	1.20	0.53-2.74
37-41 (reference)	1.00		1.00		1.00
≥42	0.91	0.30-2.85	1.01	0.61-1.66	0.67	0.13-3.84
Adjusted hazard ratios†
Gestational age (weeks)
30-33	8.01	5.18-12.38	1.91	1.27-2.86	3.74	1.42-10.23
34-36	3.39	2.43-4.72	1.25	0.98-1.59	1.21	0.51-2.89
37-41 (reference)	1.00		1.00		1.00
≥42	0.91	0.30-2.84	1.02	0.62-1.68	0.73	0.14-4.06

Back to Article Outline

Results 

Analysis Dataset 

A total of 142 735 children born at ≥30 weeks gestation were included in the initial dataset. Of these, 144 records were excluded because the infant died during the birth hospitalization, 12 records were excluded because of missing sex, 1 record with a clearly erroneous birth weight was excluded, and 1257 records were excluded because the infant did not have at least 1 day of follow-up time. Follow-up time was defined as the date of the last visit identified in the medical record minus the birth date. The distribution of follow-up time for the CP, DD/MR, and seizure groups was statistically different (P < .001) from the group without any of these conditions identified (Appendix I). The distribution of follow-up time in children with adverse outcomes was skewed toward a longer duration of time. After exclusions, 141 321 children remained for analysis.

Maternal and Infant Characteristics 

Of the 141 321 children in the study population, most were born at term (37-41 weeks; Table I). The racial/ethnic composition of mothers was similar to that in earlier reports from the KPMCP.²⁰, ²¹, ²², ³¹ For preterm births, more than half of those at 30 to 33 weeks and one third of those at 34 to 36 weeks were delivered via cesarean section. The rate of cesarean section declined further for term babies, but then increased again for post-term births (≥42 weeks). Approximately 3% of children in the study population were multiple births (twin, triplet, or higher order). The proportion of multiple births decreased with increasing gestational age. Low birth weight (<2500 grams) affected 5.3% of the cohort, with 0.4% being very low birth weight (<1500 grams). Nearly 3% of the cohort was SGA and about 21% was LGA. Approximately 1% of children had an Apgar score <7 at 5 minutes.

Table I. Number and percentage of children by maternal and infant characteristics and gestational age

	Gestational age (weeks)
	Total		30-33		34-36		37-41		42+
	Number	Percent	Number	Percent	Number	Percent	Number	Percent	Number	Percent
Total	141321	100.0	1921	1.4	8341	5.9	128955	91.2	2104	1.5
Maternal race/ethnicity
Hispanic	34557	24.5	386	20.1	1845	22.1	31828	24.7	498	23.7
Black	10332	7.3	195	10.2	744	8.9	9232	7.2	161	7.7
Asian	25723	18.2	334	17.4	1529	18.3	23598	18.3	262	12.5
White	58664	41.5	785	40.9	3397	40.7	53484	41.5	998	47.4
Other/unknown	12045	8.5	221	11.5	826	9.9	10813	8.4	185	8.8
Maternal age (years)
<20	8413	6.0	115	6.0	515	6.2	7643	5.9	140	6.7
20-29	64788	45.8	748	38.9	3434	41.2	59637	46.2	969	46.1
30-39	62421	44.2	927	48.3	3877	46.5	56718	44.0	899	42.7
≥40	5422	3.8	126	6.6	488	5.9	4716	3.7	92	4.4
Unknown/missing	277	0.2	5	0.3	27	0.3	241	0.2	4	0.2
Mode of delivery
Vaginal	108999	77.1	839	43.7	5469	65.6	101299	78.6	1392	66.2
Cesarean section	31296	22.1	1051	54.7	2776	33.3	26775	20.8	694	33.0
Unknown/missing	1026	0.7	31	1.6	96	1.2	881	0.7	18	0.9
Infant sex
Male	72277	51.1	1032	53.7	4535	54.4	65585	50.9	1125	53.5
Female	69044	48.9	889	46.3	3806	45.6	63370	49.1	979	46.5
Multiple gestation
Yes	3790	2.7	533	27.7	1412	16.9	1843	1.4	2	0.1
No	137531	97.3	1388	72.3	6929	83.1	127112	98.6	2102	99.9
Birthweight (grams)
<1500	531	0.4	457	23.8	66	0.8	8	0.0	—	0.0
<2500	7434	5.3	1810	94.2	3314	39.7	2309	1.8	1	0.0
≥2500	133887	94.7	111	5.8	5027	60.3	126646	98.2	2103	100.0
SGA (10th percentile)
Yes	3917	2.8	171	8.9	508	6.1	3209	2.5	29	1.4
No	137404	97.2	1750	91.1	7833	93.9	125746	97.5	2075	98.6
LGA (90th percentile)
Yes	30245	21.4	74	3.9	838	10.0	28720	22.3	613	29.1
No	111076	78.6	1847	96.1	7503	90.0	100235	77.7	1491	70.9
Agar score at 5 minutes
<7	1857	1.3	106	5.5	194	2.3	1532	1.2	25	1.2
≥7	139464	98.7	1815	94.5	8147	97.7	127423	98.8	2079	98.8

Race categories are non-Hispanic.

Distribution of Neurological Outcomes by Cohort Characteristics 

CP was diagnosed in 2.5 per 1000 children (Table II). An additional 9.5 per 1000 children in whom CP was not diagnosed had a diagnosis of DD/MR, and 1.3 per 1000 children with seizures did not have a diagnosis of CP or DD/MR.

Table II. Number and rate^⁎ of neurological outcomes by cohort characteristics

	No CP, DD/MR, or seizures		CP		DD/MR		Seizures
	Number	Rate	Number	Rate	Number	Rate	Number	Rate
Total	139445	986.7	357	2.5	1340	9.5	179	1.3
Gestational age (weeks)
30-33	1843	959.4	34	17.7	36	18.7	8	4.2
34-36	8166	979.0	61	7.3	102	12.2	12	1.4
37-41	127359	987.6	258	2.0	1181	9.2	157	1.2
≥42	2077	987.2	4	1.9	21	10.0	2	1.0
Maternal race/ethnicity
Hispanic	34133	987.7	66	1.9	319	9.2	39	1.1
Black	10183	985.6	30	2.9	101	9.8	18	1.7
Asian	25394	987.2	47	1.8	255	9.9	27	1.0
White	57864	986.4	171	2.9	551	9.4	78	1.3
Other/unknown	11871	985.6	43	3.6	114	9.5	17	1.4
Infant sex
Male	71014	982.5	198	2.7	966	13.4	99	1.4
Female	68431	991.1	159	2.3	374	5.4	80	1.2
Multiple gestation
Yes	3708	978.4	26	6.9	53	14.0	3	0.8
No	135737	987.0	331	2.4	1287	9.4	176	1.3
Birthweight (grams)
<1500	495	932.2	19	35.8	14	26.4	3	5.6
<2500	6722	973.8	57	8.3	108	15.6	16	2.3
≥2500	132228	987.6	281	2.1	1218	9.1	160	1.2
SGA (10th percentile)
Yes	3803	970.9	33	8.4	71	18.1	10	2.6
No	135642	987.2	324	2.4	1269	9.2	169	1.2
LGA (90th percentile)
Yes	29837	986.5	65	2.1	303	10.0	40	1.3
No	109608	986.8	292	2.6	1037	9.3	139	1.3

Race categories are non-Hispanic.

Table II also shows the distribution of neurological outcomes by selected maternal and infant characteristics. Among infants with a given characteristic, the pattern was generally one of higher rates of CP, DD/MR, and seizures for the categories traditionally thought of as highest risk (eg, early gestational age, lower birth weight, SGA, multiple gestation). The Figure displays the average incidence of CP, DD/MR, and seizures and a combined category. Across the study's gestational age spectrum, the incidence of CP, DD/MR, and the presence of any condition decreases steadily between 31 and 38 weeks gestation. Thereafter, the point rates for each of these conditions become stable.

• 
  • View Large Image
  • Download to PowerPoint

• Figure. 

  Incidence of neurodevelopmental outcomes by gestational age (3-week average incidence).

Results of Manual Audit of Randomly Selected Records 

The data audit included a review of the complete electronic patient records in the KPMCP Clinical Information Presentation system, which includes diagnoses, prescription records, some physician text notes and, in some recent entries, very detailed information from the new Kaiser Permanente HealthConnect automated electronic record currently being deployed. Thus, this audit had access to more current data than that obtained in our original download.

Auditing confirmed the original diagnostic classification (ie, that none of the 20 children in the “no CP, DD/MR, and seizures” group had CP, DD/MR, or seizures). Of the 20 randomly selected cases with CP, presence of the condition was unequivocal in 15 cases. In 1 case, the CP was secondary to traumatic injury, and the condition appeared to have resolved in 3 cases, including 1 case of resolution after successful treatment of a brain tumor. One of the audited cases had a single CP diagnosis at 7 months of age with no further mention of the problem, suggesting that the condition had resolved.

Of the 20 records with DD/MR, the diagnosis was confirmed in 17 cases, although 1 case appeared to be an instance of speech delay that the treating physician attributed to probable residual effects of temporary hearing loss caused by otitis media, and another case appeared to be improving with time. In the remaining 3 cases, the DD appeared to have resolved. Of the 20 records audited in the seizure group, 10 were confirmed for a persistent seizure disorder. Of the remaining 10 cases, 6 had seizure disorders that appeared to have resolved (2 being unusually severe recurrent febrile seizures), 2 had other conditions that might be confused with a seizure disorder (eg, paroxysmal shaking labeled as a seizure disorder but subsequently found to have a normal results on electroencephalogram), and 2 cases, both with ICD code 780.39, did not have any evidence for a seizure disorder except the presence of the indicated ICD code.

With sensitivity analyses (as aforementioned), we found that even after random reassignment of cases, the resulting mean hazard ratios remained significant and the directionality of the findings were not affected (Appendix II).

Increasing Risk of Adverse Outcome with Decreasing Gestational Age 

Table III provides crude and adjusted hazard ratios and 95% confidence intervals (CI) for CP, DD/MR, and seizures as compared with term infants. The crude hazard ratios for CP decreased with increasing gestational age, with a statistically significant association for children born at 30 to 36 weeks gestation. The hazard ratios remained statistically significant after adjustment. Compared with infants born at term, CP was approximately 8 times more likely to be diagnosed in children born between 30 and 33 weeks gestation (95% CI, 5.38-11.51) and 3 times more likely to be diagnosed in children born late preterm (95% CI, 2.54-4.52).

Table III. Cox proportional hazards ratios for neurological outcomes by gestational age

	CP		DD/MR		Seizures
	Hazard ratio	95% CI	Hazard ratio	95% CI	Hazard ratio	95% CI
Crude hazard ratios
Gestational age (weeks)
30-33	9.09	6.36-12.99	2.17	1.56-3.03	3.56	1.75-7.25
34-36	3.68	2.79-4.87	1.36	1.11-1.66	1.19	0.66-2.15
37-41 (reference)	1.00		1.00		1.00
≥42	0.91	0.34-2.46	1.02	0.66-1.56	0.74	0.19-3.00
Adjusted hazard ratios⁎
Gestational age (weeks)
30-33	7.87	5.38-11.51	1.90	1.34-2.71	3.92	1.95-7.87
34-36	3.39	2.54-4.52	1.25	1.01-1.54	1.27	0.69-2.32
37-41 (reference)	1.00		1.00		1.00
≥42	0.90	0.34-2.43	1.01	0.66-1.55	0.73	0.18-2.95

There were significantly higher risks of DD/MR for children born at 30 to 33 weeks (95% CI, 1.34-2.71) and marginally higher risks for children born at 34 to 36 weeks (95% CI, 1.01-1.54) compared with children born at term (Table III).

For seizures, a stronger effect was revealed after adjustment. However, in both models, although the hazard ratios decreased with increasing gestational age for all categories, the association was statistically significant only for the 30 to 33 week group.

Two additional sensitivity analyses were performed: 1 limited to singletons only, and another calculating outcomes for all children without using the hierarchy. Restricting the analyses to singletons had no substantial impact on the results. For analyses without the hierarchy, the calculation for rates of CP remained the same, but differed for DD/MR and seizures. The results show an additional 161 cases of DD/MR (which also had CP) and an additional 145 cases of seizures (which also had CP or DD/MR). As an example, using this revised scenario, the adjusted point estimate for DD/MR in late preterm infants increases from 1.25 to 1.47. Detailed results are available in the Appendices (available at www.jpeds.com).

Back to Article Outline

Discussion 

On the basis of a contemporary dataset from which children born extremely premature (<30 weeks) were excluded, this study demonstrates that moderate prematurity confers an increased risk for CP and DD/MR beyond infancy. Most important, we demonstrate a 3-fold increased risk of CP and significantly higher rates of DD/MR for late preterm infants compared with term infants. A total of 2.1% of the late preterm infants in our study population were identified as having CP, DD/MR, or seizures compared with 1.2% of infants born at term.

Earlier studies of CP by gestational age from the 1980s and 1990s have focused on the significantly higher rates of CP in very preterm babies.² Those studies also reported overall higher rates of CP in moderately preterm infants (different groupings between 32 and 36 weeks gestation) than term infants.³², ³³ The Institute of Medicine report on preterm birth noted that infants born between 32 and 36 weeks gestation account for 8% to 9% of live births, whereas representing 16% to 20% of children with CP.², ³³, ³⁴ Although very preterm births are at highest risk for severe adverse outcomes related to their extreme prematurity, the implications of the much more common occurrence of late preterm birth on pediatric neurodevelopmental outcomes may be under-appreciated. Our study demonstrates the increased incidence of CP and DD/MR for children born moderately preterm and provides the framework for studies pursuing the mechanisms for this association. A recent study by Chyi et al, showing worse school performance in late preterm infants, is consistent with our findings.²⁶

Late preterm births constitute 9% of all live births, more than 70% of preterm births in the United States and two-thirds of the increase in preterm birth in the last decade.⁶ Late preterm infants may be considered by both obstetric and pediatric practices to be medically similar to term infants. However, our data and the growing literature on morbidity and mortality in late preterm infants suggests otherwise: that late preterm birth is strongly associated with neurodevelopmental morbidity. In particular, the observed relationship between gestational age and neurodevelopmental outcomes is all the more striking when our study population, which has several favorable key indicators of health compared with overall U.S. births, is considered. First, our cohort was comprised of substantially heavier babies, with only 5.3% being low birth weight, compared with 8.1% for the U.S. total.³ Further, only 2.8% of our births were SGA, compared with 4.7% nationally.³⁵ Second, the children in our study, in contrast to many others in the United States had the advantage of comprehensive and coordinated health insurance coverage through KPMCP, removing at least the financial barrier of access to health care and thus appropriate preventive care, diagnostic tests, and treatments. These characteristics suggest that our findings are conservative and may underestimate the true incidence of adverse neurodevelopmental outcomes in late preterm births in the United States.

The strengths of this study include use of a large, robust, and comprehensive clinical database that allowed for the inpatient and outpatient follow-up of infants for as long as 5.5 years postpartum.³¹, ³⁶, ³⁷, ³⁸ The use of hazard ratios in the analysis permitted adequate control for different follow-up times for the individuals in the analysis. The identified rate of CP was consistent with the most current published population-based estimate of 3.1 per 1000 children in 2000.³⁹ An analysis conducted for the Institute of Medicine report on preterm birth² indicated that the overall rate of CP was 2.2 per 1000 children, but this analysis only included survivors to age 3 years. Further, that analysis included gestational age-specific rates, and although the categories overlapped with our analysis (29-32 weeks, 33-36 weeks, >37 weeks), the rates were fairly consistent. In addition, our study used a hierarchy to calculate adverse neurological outcomes. We have provided evidence showing this to be a conservative analytic approach that likely underestimates the impact of gestational age on neurological development.

Another strength is the availability of comparison groups of different gestational ages. Neurodevelopmental outcomes for babies born at <30 weeks gestation were excluded from the analysis, because this risk factor is already well-documented. Here, infants born at 30 to 33 weeks were included to serve as a comparison group for the analysis and allowed us to more fully analyze the progression of weekly differences in the neurodevelopmental outcomes of interest.

The assessment of duration of follow-up time showed a different distribution for each diagnostic outcome group (Appendix I), with longer follow-up for children with adverse outcomes than children without adverse outcomes. This may have occurred because sick children are more likely to stay in the KCMCP system longer and therefore may be more likely to receive a diagnosis of 1 of the adverse outcomes. Conversely, children without developmental delay may not have sufficient time in the system for it to be diagnosed. Cox proportional hazard models were used to adjust for the differential follow-up.

The identification of diagnoses was limited to the use of ICD-9-CM codes, therefore some individuals with neurological conditions may have been missed or incorrectly diagnosed. However, we augmented the analysis with an audit of cases, thereby providing assurance that the codes adequately identified cases. The audit indicated that a small proportion of the diagnoses resolved with time; however, this finding did not change the initial diagnosis or impact the overall robustness of the data. Moreover, the resolution of the condition does not necessarily invalidate the original diagnosis. Other studies have documented changes in diagnoses for more immature infants.⁴⁰ All cases of the selected adverse neurodevelopmental outcomes were included, not just those conditions associated with prematurity. However, these data strongly suggest that late preterm birth is associated with increased risk of the adverse neurodevelopmental outcomes of CP and DD/MR.

The basis for the increased risk of adverse neurodevelopmental outcomes in late preterm infants is unclear. One alternative is that brain growth during this vulnerable period is altered or damaged by preterm delivery per se or by complications of preterm delivery. A substantial literature documents such risk factors in very preterm infants, but the existing data suggest that these complications are uncommon for late preterm infants.³¹ Another alternative is that the events that led to the preterm delivery also contribute to the increased risk of poorer outcomes. Our current dataset was limited to neonatal ICD codes and did not include maternal or fetal indications for delivery, which are difficult to capture, or other important maternal factors, such a parity, education, smoking, alcohol use, assisted fertility therapies, antenatal steroid use, and induction of labor. Thus, we were unable to examine such issues and could not distinguish the adverse effects of early delivery from those associated with causes of early delivery, nor could we include mode of delivery as a predictor in our models. The use of assisted fertility therapies have been associated with poorer outcomes largely through the increase in the number of multiple births, but also occurring among singletons.² Finally, congenital and chromosomal anomalies may play a role, but the lack of standard nosology for categorizing anomalies for their relationship to development precludes the use of the individual ICD codes in the analysis.

In conclusion, the findings of this study indicate that there is a steadily decreasing risk of adverse neurological consequences with increasing gestational age, extending to term. These results may underestimate the full burden of neurdevelopmental disability associated with late preterm birth, because most of these children have not yet entered school when learning disabilities are more likely to be seen.²⁶ These findings highlight the need to assess outcomes of moderate and late preterm birth well beyond the neonatal period, through school years. Because brain development continues throughout the first year and preterm infants may demonstrate post-partum catch-up brain development,⁴¹ our results suggest that late preterm infants could benefit from neurological assessment and perhaps even developmental intervention. The large and growing number of late preterm infants substantiates the importance of understanding the implications of every additional gestational week for the developing child. Appendix III, Appendix IV.

Appendix III. Cox proportional hazard ratios for neurological outcomes by gestational age, singleton births only

	CP				DD/MR				Seizures
	Original results		Results limited to singletons⁎		Original results		Results limited to singletons†		Original results		Results limited to singletons‡
	Hazard ratio	95% CI	Hazard ratio	95% CI	Hazard ratio	95% CI	Hazard ratio	95% CI	Hazard ratio	95% CI	Hazard ratio	95% CI
Crude hazard ratios
Gestational age (weeks)
30-33	9.09	6.36-12.99	8.86	5.83-13.45	2.17	1.56-3.03	2.50	1.74-3.59	3.56	1.75-7.25	3.67	1.62-8.29
34-36	3.68	2.79-4.87	3.47	2.55-4.72	1.36	1.11-1.66	1.34	1.07-1.67	1.19	0.66-2.15	1.42	0.79-2.56
37-41 (reference)	1.00		1.00		1.00		1.00		1.00		1.00
≥42	0.91	0.34-2.46	0.91	0.34-2.45	1.02	0.66-1.56	1.03	0.67-1.58	0.74	0.19-3.00	0.74	0.18-2.98

Cases with no condition would decrease by 3708 from 139 445 to 135 737.

Appendix IV. Comparison of results with original hierarchy of outcomes versus all children with respective outcomes

	DD/MR				Seizures
	Original results		Results including CP cases		Original results		Results including CP and DD/MR cases
	Hazard ratio	95% CI	Hazard ratio	95% CI	Hazard ratio	95% CI	Hazard ratio	95% CI
Crude hazard ratios
Gestational age (weeks)
30-33	2.17	1.56-3.03	2.78	2.10-3.68	3.56	1.75-7.25	3.33	1.91-5.81
34-36	1.36	1.11-1.66	1.60	1.34-1.91	1.19	0.66-2.15	2.01	1.41-2.86
37-41 (reference)	1.00		1.00		1.00		1.00
≥42	1.02	0.66-1.56	0.97	0.64-1.48	0.74	0.19-3.00	0.86	0.32-2.32
Adjusted hazard ratios⁎
Gestational age (weeks)
30-33	1.90	1.34-2.71	2.45	1.82-3.29	3.92	1.95-7.87	3.23	1.82-5.72
34-36	1.25	1.01-1.54	1.47	1.22-1.77	1.27	0.69-2.32	1.98	1.38-2.82
37-41 (reference)	1.00		1.00		1.00		1.00
≥42	1.01	0.66-1.55	0.97	0.63-1.47	0.73	0.18-2.95	0.85	0.32-2.28

Back to Article Outline

We are grateful to Bruce Folck, Soora Wi, and John D. Greene of the Division of Research Perinatal Research Unit for assistance with preparing the initial dataset and for providing consultation with programming. We also thank Eileen Walsh and Myesha Smith for their assistance with designing the audit form and performing the audit, and Dr Joseph Selby and Michael J. Davidoff for reviewing the manuscript and providing specific suggestions for displaying the data, and Tomoko Yamada-Kushnir for her assistance in both reviewing and revising the manuscript.

Back to Article Outline

References 

1. Mathews TJ, MacDorman MF. Infant mortality statistics from the 2004 period linked birth/infant death data set. Natl Vital Stat Rep. 2007;55:1–32
   • View In Article
   • MEDLINE
2. Institute of Medicine, Committee on Understanding Premature Births and Assuring Healthy Outcomes, Board on Health Sciences Policy. Preterm birth. Causes, consequences, and prevention. Washington, D.C: National Academies Press; 2007;
   • View In Article
3. Martin JA, Hamilton BE, Sutton PD, Ventura SJ, Menacker F, Kirmeyer S. Births: final data for 2004. Natl Vital Stat Rep. 2006;55:1–101
   • View In Article
   • MEDLINE
4. Engle WA. A recommendation for the definition of “late preterm” (near-term) and the birth weight-gestational age classification system. Semin Perinatol. 2006;30:2–7
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (230 KB)
   • CrossRef
5. Engle WA, Tomashek KM, Wallman C. “Late-preterm” infants: a population at risk. Pediatrics. 2007;120:1390–1401
   • View In Article
   • CrossRef
6. Davidoff MJ, Dias T, Damus K, Russell R, Bettegowda VR, Dolan S, et al. Changes in the gestational age distribution among U.S. singleton births: impact on rates of late preterm birth, 1992 to 2002. Semin Perinatol. 2006;30:8–15
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (342 KB)
   • CrossRef
7. Hack M, Fanaroff AA. Outcomes of children of extremely low birthweight and gestational age in the 1990s. Semin Neonatol. 2000;5:89–106
   • View In Article
   • Abstract
   • Full-Text PDF (217 KB)
   • CrossRef
8. Vohr BR, Wright LL, Poole WK, McDonald SA. Neurodevelopmental outcomes of extremely low birth weight infants <32 weeks' gestation between 1993 and 1998. Pediatrics. 2005;116:635–643
   • View In Article
   • CrossRef
9. Robertson CM, Watt MJ, Yasui Y. Changes in the prevalence of cerebral palsy for children born very prematurely within a population-based program over 30 years. JAMA. 2007;297:2733–2740
   • View In Article
   • CrossRef
10. Kramer MS, Demissie K, Yang H, Platt RW, Sauve R, Liston R. The contribution of mild and moderate preterm birth to infant mortality (Fetal and Infant Health Study Group of the Canadian Perinatal Surveillance System). JAMA. 2000;284:843–849
    • View In Article
    • MEDLINE
    • CrossRef
11. Young PC, Glasgow TS, Li X, Guest-Warnick G, Stoddard G. Mortality of late-preterm (near-term) newborns in Utah. Pediatrics. 2007;119:e659–e665
    • View In Article
    • CrossRef
12. Tomashek KM, Shapiro-Mendoza CK, Davidoff MJ, Petrini JR. Differences in mortality between late-preterm and term singleton infants in the United States, 1995-2002. J Pediatr. 2007;151:450–456
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (154 KB)
    • CrossRef
13. Raju TN, Higgins RD, Stark AR, Leveno KJ. Optimizing care and outcome for late-preterm (near-term) infants: a summary of the workshop sponsored by the National Institute of Child Health and Human Development. Pediatrics. 2006;118:1207–1214
    • View In Article
    • CrossRef
14. Raju TN. Epidemiology of late preterm (near-term) births. Clin Perinatol. 2006;33:751–763abstract vii
    • View In Article
    • Full Text
    • Full-Text PDF (300 KB)
    • CrossRef
15. Wang ML, Dorer DJ, Fleming MP, Catlin EA. Clinical outcomes of near-term infants. Pediatrics. 2004;114:372–376
    • View In Article
    • CrossRef
16. Sarici SU, Serdar MA, Korkmaz A, Erdem G, Oran O, Tekinalp G, et al. Incidence, course, and prediction of hyperbilirubinemia in near-term and term newborns. Pediatrics. 2004;113:775–780
    • View In Article
    • CrossRef
17. Escobar GJ, Clark RH, Greene JD. Short-term outcomes of infants born at 35 and 36 weeks gestation: we need to ask more questions. Semin Perinatol. 2006;30:28–33
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (106 KB)
    • CrossRef
18. McCormick MC, Escobar GJ, Zheng Z, Richardson DK. Place of birth and variations in management of late preterm (“near-term”) infants. Semin Perinatol. 2006;30:44–47
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (92 KB)
    • CrossRef
19. Tomashek KM, Shapiro-Mendoza CK, Weiss J, Kotelchuck M, Barfield W, Evans S, et al. Early discharge among late preterm and term newborns and risk of neonatal morbidity. Semin Perinatol. 2006;30:61–68
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (118 KB)
    • CrossRef
20. Escobar GJ, Joffe S, Gardner MN, Armstrong MA, Folck BF, Carpenter DM. Rehospitalization in the first two weeks after discharge from the neonatal intensive care unit. Pediatrics. 1999;104:e2
    • View In Article
    • CrossRef
21. Escobar GJ, Gonzales VM, Armstrong MA, Folck BF, Xiong B, Newman TB. Rehospitalization for neonatal dehydration: a nested case-control study. Arch Pediatr Adolesc Med. 2002;156:155–161
    • View In Article
    • MEDLINE
22. Escobar GJ, Greene JD, Hulac P, Kincannon E, Bischoff K, Gardner MN, et al. Rehospitalization after birth hospitalization: patterns among infants of all gestations. Arch Dis Child. 2005;90:125–131
    • View In Article
    • CrossRef
23. Escobar GJ, McCormick MC, Zupancic JA, Coleman-Phox K, Armstrong MA, Greene JD, et al. Unstudied infants: outcomes of moderately premature infants in the neonatal intensive care unit. Arch Dis Child Fetal Neonatal Ed. 2006;91:F238–F244
    • View In Article
    • MEDLINE
    • CrossRef
24. Gilbert WM, Nesbitt TS, Danielsen B. The cost of prematurity: quantification by gestational age and birth weight. Obstet Gynecol. 2003;102:488–492
    • View In Article
    • MEDLINE
    • CrossRef
25. Raju TN. The problem of late-preterm (near-term) births: a workshop summary. Pediatr Res. 2006;60:775–776
    • View In Article
    • MEDLINE
    • CrossRef
26. Chyi LJ, Lee HC, Hintz SR, Gould JB, Sutcliffe TL. School outcomes of late preterm infants: special needs and challenges for infants born at 32 to 36 weeks gestation. J Pediatr. 2008;153:25–31
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (117 KB)
    • CrossRef
27. Escobar GJ, Fischer A, Kremers R, Usatin MS, Macedo AM, Gardner MN. Rapid retrieval of neonatal outcomes data: the Kaiser Permanente Neonatal Minimum Data Set. Qual Manag Health Care. 1997;5:19–33
    • View In Article
    • MEDLINE
28. Go AS, Hylek EM, Chang Y, Phillips KA, Henault LE, Capra AM, et al. Anticoagulation therapy for stroke prevention in atrial fibrillation: how well do randomized trials translate into clinical practice?. JAMA. 2003;290:2685–2692
    • View In Article
    • CrossRef
29. Wu YW, March WM, Croen LA, Grether JK, Escobar GJ, Newman TB. Perinatal stroke in children with motor impairment: a population-based study. Pediatrics. 2004;114:612–619
    • View In Article
    • CrossRef
30. Last JM, Abramson JH, Friedman GD, Porta M, Spasoff RA, Thuriaus M. A dictionary of epidemiology. In: 3rd ed. New York, NY: Oxford University Press; 1995;p. 108
    • View In Article
31. Wu YW, Escobar GJ, Grether JK, Croen LA, Greene JD, Newman TB. Chorioamnionitis and cerebral palsy in term and near-term infants. JAMA. 2003;290:2677–2684
    • View In Article
    • CrossRef
32. Cummins SK, Nelson KB, Grether JK, Velie EM. Cerebral palsy in four northern California counties, births 1983 through 1985. J Pediatr. 1993;123:230–237
    • View In Article
    • Abstract
    • Full-Text PDF (672 KB)
    • CrossRef
33. Hagberg B, Hagberg G, Beckung E, Uvebrant P. Changing panorama of cerebral palsy in Sweden. VIII (Prevalence and origin in the birth year period 1991-94). Acta Paediatr. 2001;90:271–277
    • View In Article
    • MEDLINE
    • CrossRef
34. MacGillivray I, Campbell DM. The changing pattern of cerebral palsy in Avon. Paediatr Perinat Epidemiol. 1995;9:146–155
    • View In Article
    • MEDLINE
    • CrossRef
35. March of Dimes, Perinatal Data Center. Unpublished statistics (Data Source: NCHS 2004 Natality File). In; 2007.
    • View In Article
36. Escobar GJ, Liljestrand P, Hudes ES, Ferriero DM, Wu YW, Jeremy RJ, et al. Five-year neurodevelopmental outcome of neonatal dehydration. J Pediatr. 2007;151:127–133
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (191 KB)
    • CrossRef
37. Lee J, Croen LA, Backstrand KH, Yoshida CK, Henning LH, Lindan C, et al. Maternal and infant characteristics associated with perinatal arterial stroke in the infant. JAMA. 2005;293:723–729
    • View In Article
    • CrossRef
38. Newman TB, Liljestrand P, Jeremy RJ, Ferriero DM, Wu YW, Hudes ES, et al. Outcomes among newborns with total serum bilirubin levels of 25 mg per deciliter or more. N Engl J Med. 2006;354:1889–1900
    • View In Article
    • CrossRef
39. Bhasin TK, Brocksen S, Avchen RN, Van Naarden Braun K. Prevalence of four developmental disabilities among children aged 8 years—Metropolitan Atlanta Developmental Disabilities Surveillance Program, 1996 and 2000. MMWR Surveill Summ. 2006;55:1–9
    • View In Article
40. Kitchen WH, Ford GW, Rickards AL, Lissenden JV, Ryan MM. Children of birth weight less than 1000 g: changing outcome between ages 2 and 5 years. J Pediatr. 1987;110:283–288
    • View In Article
    • Abstract
    • Full-Text PDF (483 KB)
    • CrossRef
41. National Research Council and Institute of Medicine. From neurons to neighborhoods: the science of early childhood development. In:  Shonkoff JP,  Phillips DA editor. Chapter 8. Washington, DC: National Academy Press; 2000;p. 206–212
    • View In Article