Late Preterm Infants: Birth Outcomes and Health Care
Utilization in the First Year
WHAT’S KNOWN ON THIS SUBJECT: LPIs have been shown to be at greater risk of poor outcomes than term infants, but studies often include infants with birth defects or multiple births, who are at greater risk of both preterm birth and poor outcomes.
WHAT THIS STUDY ADDS: After exclusion of infants with birth defects, multiple births, and with propensity score matching, LPIs are at greater risk of a wide range of morbidities, as well as higher inpatient and outpatient costs, in the first year of life.
abstract
OBJECTIVE:To distinguish the effects of late preterm birth from the complications associated with the causes of delivery timing, this study used propensity score–matching methods on a statewide database that contains information on both mothers and infants.
METHODS:Data for this study came from Arkansas Medicaid claims data linked to state birth certificate data for the years 2001 through 2005. We excluded all multiple births, infants with birth defects, and infants at⬍33 weeks of gestation. Late preterm infants (LPIs) (34 to 36 weeks of gestation) were matched with term infants (37– 42 weeks of gestation) according to propensity scores, on the basis of infant, ma-ternal, and clinical characteristics.
RESULTS:A total of 5188 LPIs were matched successfully with 15 303 term infants. LPIs had increased odds of poor outcomes during their birth hospitalization, including a need for mechanical ventilation (ad-justed odds ratio [aOR]: 1.31 [95% confidence interval [CI]: 1.01–1.68]), respiratory distress syndrome (aOR: 2.84 [95% CI: 2.33–3.45]), and hy-poglycemia (aOR: 1.60 [95% CI: 1.26 –2.03]). Outpatient and inpatient Medicaid expenditures in the first year were both modestly higher (outpatient, adjusted marginal effect: $108 [95% CI: $58 –$158]; inpa-tient, $597 [95% CI: $528 –$666]) for LPIs.
CONCLUSIONS:LPIs are at increased risk of poor health-related out-comes during their birth hospitalization and of increased health care utilization during their first year.Pediatrics2010;126:e311–e319
AUTHORS:T. Mac Bird, MS,a,bJanet M. Bronstein, PhD,c
Richard W. Hall, MD,aCurtis L. Lowery, MD,dRichard
Nugent, MD, MPH,eand Glen P. Mays, PhD, MPHb
Departments ofaPediatrics,bHealthcare Policy and
Management, anddObstetrics and Gynecology, University of
Arkansas for Medical Sciences, Little Rock, Arkansas;
cDepartment of Health Care Organization and Policy, School of
Public Health, University of Alabama at Birmingham, Birmingham, Alabama; andeCenter for Local Public Health,
Arkansas Department of Health, Little Rock, Arkansas
KEY WORDS
late preterm, neonatal outcomes, health care utilization, Medicaid, propensity score
ABBREVIATIONS OR— odds ratio aOR—adjusted odds ratio CI— confidence interval LPI—late preterm infant
www.pediatrics.org/cgi/doi/10.1542/peds.2009-2869
doi:10.1542/peds.2009-2869
Accepted for publication Apr 15, 2010
Address correspondence to T. Mac Bird, MS, University of Arkansas for Medical Sciences, Department of Pediatrics, Center for Applied Research and Evaluation, 1 Children’s Way, Slot 512-26, Little Rock, AR 72202-3591. E-mail: birdtommym@uams.edu
PEDIATRICS (ISSN Numbers: Print, 0031-4005; Online, 1098-4275).
Copyright © 2010 by the American Academy of Pediatrics
The proportion of preterm births in the United States increased from 10.5% in 1990 to 12.6% of all births in 2005, a 20% increase.1Late preterm births, at
340⁄7 to 366⁄7
weeks of gestation, repre-sent nearly 75% of all preterm births and constitute approximately two-thirds of the recent rate increase.1
Many late preterm infants (LPIs) are similar in size to term infants and therefore may be treated by caregiv-ers and health care professionals as if they are developmentally similar to term infants.2,3However, LPIs are
phys-iologically immature and seem to be at greater risk of morbidity and death than are term infants.2
Many obstetric decisions during the fi-nal weeks of a pregnancy involve weighing the risks and benefits of de-livering the infant prematurely against the risks and benefits of extending the pregnancy.4,5 For fully informed
decision-making, an accurate under-standing of the risks related to either choice is necessary. However, most outcomes research involving prema-ture infants has focused on infants born at⬍34 weeks of gestation.6–9This
is true although the public health im-pact of late preterm births, because of their large numbers, is potentially as great as or greater than that of early or moderate preterm births.10
There is a small but growing body of research documenting increased risks of morbidity and death in the neo-natal period for LPIs, compared with term infants. The research includes studies of complications during the birth hospitalization,7,11–18
rehospital-ization rates,12,17,19–21 risk factors for
morbidity,2,18,20,22–26 mortality rates,11,21,27,28
and long-term outcomes.8,9,29–34Those
studies showed substantial increases in the risks of morbidity and death for LPIs, compared with term infants.
Identification of the outcomes attribut-able to late preterm birth is compli-cated by the fact that a variety of
pre-natal conditions, maternal risks, and clinical practices may influence both preterm birth and subsequent adverse outcomes. The extent to which gesta-tional age is associated independently with adverse outcomes, after account-ing for these confoundaccount-ing factors, re-mains undetermined. Without control-ling for these confounding factors, it is not clear whether a LPI born after an otherwise-uncomplicated pregnancy is at any greater risk of morbidity than is a term infant. To address these questions, we limited our analysis to a cohort of relatively healthy infants. We also used a propensity score– matching method to adjust for covari-ates likely to confound the association between late preterm delivery and birth outcomes. Propensity score matching is a method of balancing ob-served characteristics that reduces selection bias and strengthens causal inferences in observational stud-ies.35,36 Although propensity score
matching cannot account for bias that results from unobserved characteris-tics, the resulting estimates can im-prove on those based on naive regres-sion models, increasing our insights into the health consequences of late preterm deliveries.
METHODS
Database
The data evaluated in this investigation were from matched birth certificates, maternal and infant Medicaid claims, and maternal and infant hospital dis-charge records for all Medicaid deliv-eries in the state of Arkansas between April 1, 2001, and December 31, 2005. This linked data set was created in co-operation with the Arkansas Division of Medical Assistance and the Arkan-sas Department of Health. During this period, Medicaid was the primary payer for 50% to 55% of the births in Arkansas. Maternal Medicaid claims data and birth certificate data
were probabilistically matched,37which
yielded a population of 91 902 Medicaid-covered deliveries. Of those cases, 73 359 (79.8%) were matched to infant claims for the first year after birth. The remaining maternal cases had insuffi-cient identifying information for match-ing to the infant claims or did not gener-ate infant claims because of early death or lack of Medicaid eligibility for the in-fant. Match rates were higher for infants born at 35 to 36 weeks of gestation (80.9%) or ⱖ37 weeks of gestation (83.2%) than for less-mature infants. This study was approved by the institu-tional review board of the University of Arkansas for Medical Sciences.
Subjects
All births in Arkansas between April 1, 2001, and December 31, 2005, with Medicaid as the primary payer, for which infant claims were matched to birth certificates and maternal claims were eligible for inclusion. In an at-tempt to isolate the effects of gesta-tional age on poor health outcomes, we created a cohort of otherwise-healthy infants by excluding those with readily identifiable medical conditions that might cause both premature birth and poor outcomes.20Excluded infants
included all infants born at ⬍340⁄7 weeks of gestation, all infants of un-known gestational age, all infants born at⬎426⁄7
weeks of gestation, all infants born weighing ⬍1500 g, all infants from multiple births, and all infants with a birth defect documented on the birth certificate or an International Classification of Diseases, Ninth Revi-sion, Clinical Modification code from the infant hospital discharge record for any of 35 major birth defects.38,39
Gestational ages were estimated from birth certificate data.
Analysis
Infants were considered either LPIs (340⁄7 to 366⁄7 weeks of gestation) or
ges-tation). Outcome measures that were based on International Classification of Diseases, Ninth Revision, Clinical Modification diagnosis codes from the birth hospitalization included indica-tor variables for sepsis (codes 038 and 771.81), respiratory distress syn-drome (code 769), transitory tachy-pnea of the newborn (code 770.6), jaundice (codes 774.0 –774.3, 774.5– 774.6, and 782.4), hypoglycemia (code 775.6), temperature instability (codes 778.2–778.4), feeding problems (code 779.3), a need for mechanical ventilation (codes 96.04 and 96.7–96.79), and apnea (code 770.8). Outcome measures that were based on linked birth and death certificate data included neonatal death (0 –28 days) and postneonatal death (29 – 365 days). Outcome measures that were based on Medicaid claims data in-cluded rehospitalization during the first year, hospital costs above $10 000 in the first year, hospital costs above $25 000 in the first year, length of stay during the birth hospitalization episode (including the birth hospitalization and any subse-quent transfer hospitalizations), total hospital days during the first year, total outpatient costs during the first year, to-tal inpatient costs during the first year (including hospital and physician-related costs), and total health care costs during the first year. All cost data indicated the actual amount paid by Medicaid for a given service or hospitalization. Control variables in-cluded indicator variables for gender, parity, gravidity, mother born in the same state as the child, maternal smok-ing, maternal drinksmok-ing, marital status, maternal age, maternal race, maternal education, maternal weight gain, inten-sity of prenatal care, cesarean delivery, induced labor, stimulated labor, vaginal birth after cesarean section, use of toco-lytic drugs, amniocentesis, birth in a hos-pital with a NICU, and 25 potential mater-nal complications, as derived from birth certificate and Medicaid claims data, in-cluding maternal fever, meconium
stain-ing, premature rupture of membranes, placental abruption and other maternal bleeding, seizures, complications of the labor process, malpresentation, cepha-lopelvic disproportion, cord prolapse, anesthetic complications, fetal distress, anemia, cardiac disease, lung disease, diabetes mellitus, herpes, hydramnios/ oligohydramnios, hemoglobinopathy, hy-pertension/eclampsia, incompetent cer-vix, previous infant weighing⬎4000 g, previous preterm birth, renal disease, Rh factor sensitization, and other (eg, co-agulation disorders, habitual aborter, or previous fetal anomaly).
LPIs were matched with term infants on the basis of propensity scores. Pro-pensity score matching is a method of balancing observed characteristics that reduces selection bias and strengthens causal inferences in observational studies.35,36 Propensity
score matching is a method of multi-variate matching that allows for close but not exact matches.40This allows for
simultaneous matching with respect to a large number of covariates in rel-atively small data sets, unlike exact matching, for which the minimal sam-ple size required increases exponen-tially with each additional covariate matched. Propensity scores were esti-mated from a logistic regression model that included all control vari-ables. In this case, the propensity score was the predicted probability, on the basis of observed variables, of a given infant being born in the late pre-term period. The resulting propensity scores were then entered into the Stata PSMATCH2 command for 3:1 matching. Matching was limited to the area of common support of the pro-pensity score, the area in which the distribution of propensity scores for LPIs overlaps with the distribution of propensity scores for term infants. This excluded infants who, on the basis of observed characteristics, were the least like those in the opposing group
and thus were least likely to produce a close match. Matches were made within a defined distance of 0.1 SD of the propensity score. The resulting matched sample was used for analysis of all outcomes. Because small differ-ences between groups remain after propensity score matching, the esti-mated propensity scores and all co-variates were included in all analyses of the matched samples.41
Logistic regression models were used to analyze all dichotomous outcomes. Negative binomial regression models were used to analyze all count out-comes. Generalized linear models us-ing a gamma distribution and a loga-rithmic link were used to analyze the highly skewed cost data.42,43Separate
regression models were calculated for each outcome variable. Crude odds ra-tios (ORs), adjusted ORs (AORs), and propensity score–matched AORs were reported for each logistic regression analysis, with 95% confidence inter-vals (CIs). For the negative binomial re-gression and generalized linear mod-els, the marginal effects in terms of days or dollars and 95% CIs were re-ported. All analyses were conducted with Stata MP 11.0 (Stata, College Station, TX).
RESULTS
A total of 5199 LPIs and 50 907 term infants met the eligibility criteria. Of those, 5188 LPIs were matched suc-cessfully to 15 303 term infants. The distributions of control variables for each group before and after propen-sity score matching are presented in Table 1. Although maternal age of⬍20 years remained statistically signifi-cant after propensity score matching, the difference between groups in the proportions of women of this age was greatly reduced.41 The proportions of
TABLE 1 Infant, Maternal, and Delivery Characteristics of LPIs and Term Infants, Before and After Propensity Score Matching
Before Propensity Score Matching After Propensity Score Matching
LPIs Term Infants P LPIs Term Infants P
Infant characteristics
N 5199 50 907 5188 15 303
Birth weight, mean, gb 2930 3311 ⬍.001a 2931 3261 ⬍.001a
Male, % 53.0 50.4 ⬍.001a 53.0 53.0 .992
Weekend birth, % 21.2 17.4 ⬍.001a 21.1 20.7 .522
Maternal characteristics
Parity 0, % 39.3 42.2 ⬍.001a 39.3 39.5 .793
Gravidity 0, % 34.8 37.6 ⬍.001a 34.9 34.8 .959
Arkansas birth, % 62.3 58.4 ⬍.001a 62.3 62.2 .891
Smoking, % 26.8 23.8 ⬍.001a 26.7 27.2 .477
Drinking, % 1.0 0.8 .097 1.0 0.9 .561
Married, % 38.4 42.7 ⬍.001a 38.4 38.1 .745
Age, %
⬍20 y 16.8 13.9 ⬍.001a 16.8 15.3 .010a
20–24 y 35.7 37.9 .002a 35.7 35.0 .399
25–29 y 16.1 17.7 .004a 16.1 16.4 .612
30–34 y 7.4 6.8 .084 7.4 7.7 .526
35–39 y 2.8 2.4 .048a 2.8 2.8 .839
ⱖ40 y 1.5 0.6 ⬍.001a 1.5 1.4 .700
Race/ethnicity, %
White 55.9 61.2 ⬍.001a 55.9 56.3 .682
Black 32.2 25.6 ⬍.001a 32.2 32.4 .728
Hispanic 10.3 11.6 .004a 10.3 9.8 .313
Other/unknown 1.6 1.6 .934 1.6 1.5 .599
Education, %
Less than high school 36.5 32.1 ⬍.001a 36.5 36.4 .940
High school 44.6 47.7 ⬍.001a 44.6 44.7 .990
More than high school 18.0 19.5 .011a 18.0 18.0 .983
Unknown 0.9 0.7 .111 0.9 0.9 .853
Weight gain, %
0–15 lb 15.0 11.7 ⬍.001a 15.0 15.2 .819
15–30 lb 41.7 40.1 .027a 41.7 40.9 .337
ⱖ31 lb 32.2 39.7 ⬍.001a 32.3 32.9 .399
Unknown 11.1 8.6 ⬍.001a 11.0 11.0 .987
Delivery characteristics
Primary cesarean section, % 13.6 15.4 .001a 13.6 14.0 .536
Secondary cesarean section, % 12.4 11.4 .026a 12.4 12.4 .984
Labor induction, % 16.5 25.4 ⬍.001a 16.5 17.4 .173
Labor stimulation, % 20.3 19.1 .036a 20.2 19.7 .474
VBAC, % 1.4 1.2 .111 1.5 1.4 .994
Tocolitic treatment, % 2.7 1.9 ⬍.001a 2.7 2.6 .556
Amniocentisis, % 0.7 0.5 .047a 0.7 0.7 .540
Prenatal care, %
None 2.2 1.3 ⬍.001a 2.2 2.3 .769
Inadequate 15.0 19.9 ⬍.001a 15.0 15.4 .555
Intermediate 27.5 43.8 ⬍.001a 27.5 28.3 .263
Adequate 38.6 23.5 ⬍.001a 38.5 37.6 .222
Intensive 11.4 9.3 ⬍.001a 11.4 11.2 .679
Unknown 5.4 2.2 ⬍.001a 5.3 5.3 .868
Complications of labor and delivery, %c
1 35.3 34.8 .429 35.4 37.5 .007a
ⱖ2 28.0 27.8 .814 27.8 27.6 .775
Hospital characteristics
Hospital with NICU, % 29.7 26.9 ⬍.001a 29.7 30.3 .453
High-volume hospital, % 43.6 44.4 .262 43.6 43.9 .725
Low-volume hospital, % 4.3 4.3 .872 4.2 4.4 .646
VBAC indicates vaginal birth after cesarean section.
aStatistically significant atP⬍.05.
bNot included in propensity score matching algorithm.
cTwenty-five potential maternal complications derived from birth certificate and Medicaid claims data, including maternal fever, meconium staining, premature rupture of membranes,
matching. This might be misleading. Complications of labor and delivery were reported in the aggregate form (ie, 1 complication versusⱖ2 compli-cations). This method of reporting saves space in the table at the expense of losing detail about the relative fre-quency of more-severe and less-severe complications. However, the 25 compli-cations on which these data were based were controlled for individually in the propensity score–matching process.
Rates of outcome variables for LPIs and term infants, as well as crude ORs before propensity score matching, are presented in Table 2. ORs for all out-come variables except neonatal death, postneonatal death, and a need for me-chanical ventilation were statistically significant and elevated for LPIs. Sta-tistically significant ORs ranged from a low of 1.23 (95% CI: 1.13–1.33) for re-hospitalization to highs of 3.21 (95% CI: 2.55– 4.04) for hospital costs above $25 000 in the first year and 3.24 (95% CI: 2.80 –3.77) for respiratory distress. The health care utilization of LPIs and term infants from the birth hospitaliza-tion through the first year is presented in Table 3. During their first year, LPIs spent 0.82 days (95% CI: 0.75– 0.88 days) more in the hospital, had $244
(95% CI: $207–$281) more in outpa-tient costs, and had $844 (95% CI: $778 –$910) more in inpatient costs, compared with term infants.
AORs from comparisons of outcomes for LPIs and term infants, before and
after propensity score matching, are presented in Table 4. Before propen-sity score matching, AORs for all out-come variables except neonatal death,
postneonatal death, and a need for me-chanical ventilation were statistically significant and elevated for LPIs. Sta-tistically significant AORs ranged from a low of 1.14 (95% CI: 1.04 –1.24) for
rehospitalization to a high of 3.20 (95% CI: 2.74 –3.74) for respiratory distress. After propensity score matching, AORs for all outcome variables except neo-natal death and postneoneo-natal death were statistically significant and
ele-vated for LPIs. AORs ranged from a low of 1.11 (95% CI: 1.01–1.23) for rehospi-talization to a high of 2.84 (95% CI: 2.33–3.45) for respiratory distress.
Adjusted marginal effects from com-parisons of health care utilization in the first year for LPIs and term infants, before and after propensity score TABLE 2 Results of Bivariate Analysis Comparing Outcomes for LPIs and Term Infants Before
Propensity Score Matching
Outcome Proportion, % OR (95% CI)
LPIs Term Infants
Death in first 28 d 0.02 0.01 1.40 (0.17–11.4)
Death in 29–365 d 0.23 0.15 1.59 (0.86–2.93)
Rehospitalized in first year 14.2 11.8 1.23 (1.13–1.33)a
Hospital costs above $10 000 in first year 8.88 3.23 2.92 (2.63–3.25)a Hospital costs above $25 000 in first year 1.81 0.57 3.21 (2.55–4.04)a
Mechanical ventilation 2.17 1.77 1.40 (0.49–3.98)
Respiratory distress 4.62 1.47 3.24 (2.80–3.77)a
Transitory tachypnea 4.70 2.47 1.95 (1.69–2.24)a
Apnea 4.41 1.93 2.35 (2.03–2.72)a
Sepsis 0.71 0.35 2.03 (1.42–2.89)a
Jaundice 17.38 9.64 1.97 (1.83–2.13)a
Hypoglycemia 2.38 1.35 1.77 (1.46–2.15)a
Temperature instability 1.11 0.58 1.92 (1.45–2.55)a
Feeding problems 4.12 1.78 2.37 (2.04–2.76)a
aStatistically significant atP⬍.05.
TABLE 3 Results of Bivariate Analysis Comparing Health Care Utilization in First Year for LPIs and Term Infants Before Propensity Score Matching
Outcome LPIs Term Infants Difference (95% CI)
Birth hospitalization length of stay, mean, d 2.61 1.96 0.64 (0.60–0.69)a Total hospital time in first year, mean, d 3.26 2.44 0.82 (0.75–0.88)a Total outpatient costs in first year, mean, $ 1560 1316 244 (207–281)a Total hospital costs in first year, mean, $ 3027 2183 844 (778–910)a Total health care costs in first year, mean, $ 4541 3472 1069 (981–1158)a
aStatistically significant atP⬍.05.
TABLE 4 Results From Multivariate Regression Models and Propensity Score–Matched Regression Models Comparing Outcomes for LPIs and Term Infants
Outcome AOR (95% CI)
Multivariate Regression Models
Propensity Score–Matched Regression Models
Death in first 28 d 1.43 (0.16–12.73) 1.38 (0.12–15.34)
Death in 29–365 d 1.51 (0.80–2.85) 1.68 (0.80–3.53)
Rehospitalized in first year 1.14 (1.04–1.24)a 1.11 (1.01–1.23)a Hospital costs above $10 000 in first year 2.48 (2.22–2.77)a 2.21 (1.93–2.53)a Hospital costs above $25 000 in first year 2.74 (2.15–3.48)a 2.04 (1.53–2.73)a Mechanical ventilation 1.23 (0.99–1.53) 1.31 (1.01–1.68)a Respiratory distress 3.20 (2.74–3.74)a 2.84 (2.33–3.45)a Transitory tachypnea 1.90 (1.64–2.19)a 1.87 (1.57–2.23)a
Apnea 2.26 (1.94–2.64)a 2.33 (1.93–2.82)a
Sepsis 2.01 (1.39–2.90)a 2.06 (1.30–3.27)a
Jaundice 1.96 (1.80–2.12)a 1.88 (1.71–2.07)a
Hypoglycemia 1.64 (1.35–2.01)a 1.60 (1.26–2.03)a
Temperature instability 1.81 (1.35–2.42)a 1.80 (1.26–2.56)a
Feeding problems 2.47 (2.11–2.89)a 2.34 (1.93–2.84)a
matching, are presented in Table 5. The propensity score–matched mod-els showed that LPIs spent 0.71 days (95% CI: 0.75– 0.88 days) more in the hospital, had $108 (95% CI: $58 –$158) more in outpatient costs, and had $597 (95% CI: $528 –$666) more in inpatient costs during their first year, compared with term infants.
DISCUSSION
With the rate of late preterm births in-creasing in the United States, deci-sions about the management of late preterm births are becoming a larger part of obstetric practice, and LPIs are using a larger proportion of the na-tion’s hospital resources. Accurate es-timates of the risks of morbidity and death associated with late preterm births are needed to enable physicians and policymakers to make informed decisions, given the changes in trends. Because of the large number of infants in this cohort, any increased morbidity might have a very large public health impact, especially if effects of these morbidities persist beyond the birth hospitalization.
Several smaller studies in the pub-lished literature correspond generally to our findings. Wang et al,13when
ex-amining 90 randomly selected LPI records and 95 randomly selected term infant records from a single insti-tution, found LPIs to be at increased risk of respiratory distress, hypoglyce-mia, feeding difficulties, jaundice, tem-perature instability, and receipt of a
sepsis evaluation. However, mechani-cal ventilation and death were not ad-dressed. Escobar et al12examined a
co-hort of ⬎47 000 infants born at 6 Kaiser Permanente hospitals between 2002 and 2004. They found that 20.6% of infants born at 33 to 34 weeks of gestation, 7.3% of infants born at 35 to 36 weeks of gestation, and only 0.6% of infants born at 37 to 42 weeks of ges-tation experienced some degree of re-spiratory distress. These rates are slightly higher than the unadjusted rates in the current study. This differ-ence in findings is likely attributable to differences in the exclusion criteria used in the current study and the num-ber of controlled maternal complica-tions. Arnon et al14 studied 207 LPIs
born at 34 to 36 weeks of gestation at a single institution in Israel between 1992 and 1998. Although those authors did not group gestational ages as in the current study, their results are similar to ours. They found that⬎5% of infants born at 34 and 35 weeks of gestation had nosocomial sepsis, com-pared with no infants born at 36 weeks of gestation; the incidence was⬍1% in our cohort. The reasons for this are unclear but may be attributable to dif-ferences in obstetric practices, includ-ing the liberal use of prenatal antibi-otic treatment in later years. The authors also found that the rate of re-spiratory distress decreased similarly as gestational age increased. Although our incidence was less than theirs (15% vs 5%), the difference was
possi-bly related to differences in defining respiratory distress and our adjust-ments for maternal factors. Gilbert et al,10using California data from 1996,
found LPIs to be at increased risk of respiratory distress and need of me-chanical ventilation, but other morbid-ities were not addressed. Khashu et al,11using population-based data from
British Columbia, found increased risks of neonatal infections and respi-ratory distress and longer lengths of hospital stay for LPIs, compared with term infants, and McLaurin et al29
found increased health care utilization during the first year for LPIs. Neither study excluded infants with birth de-fects. Melamed et al,18using data from
a single large institution in Israel, found increased rates of respiratory morbidity, infectious morbidity, jaun-dice, and hypoglycemia for LPIs, com-pared with term infants; the AORs for these conditions ranged from⬃9 to 15, whereas our corresponding AORs ranged from 1.5 to 3.
Methodologic issues might account for some of the differences in results be-tween the aforementioned studies and the current study. Many of those stud-ies did not make direct comparisons between LPIs and term infants. Most of the aforementioned studies did not ex-clude infants with birth defects, multi-ple births, or infants with birth weights of⬍1500 g. None of those studies con-trolled for as many maternal demo-graphic characteristics or complica-tions of labor and delivery as did the current study. Also, none of the afore-mentioned studies used propensity score–matching methods in an at-tempt to address selection bias. This implies that maternal complications leading to preterm labor in this popu-lation are important factors leading to neonatal complications.
Several studies found LPIs to have 2 to 5 times the risk of neonatal death as term infants.11,27,28 However, those
TABLE 5 Results From Multivariate Regression Models and Propensity Score–Matched Regression Models Comparing Health Care Utilization in First Year for LPIs and Term Infants
Outcome Adjusted Marginal Effect (95% CI)
Multivariate Regression Models
Propensity Score–Matched Regression Models
Birth hospitalization length of stay 0.57 (0.53–0.62)a 0.59 (0.54–0.64)a Total hospital time in first year 0.69 (0.63–0.75)a 0.71 (0.63–0.78)a Total outpatient costs in first year 119 (76–162)a 108 (58–158)a Total hospital costs in first year 605 (546–664)a 597 (528–666)a Total health care costs in first year 745 (664–826)a 734 (640–829)a
studies included infants with signifi-cant risk factors for both preterm birth and neonatal morbidity in their samples and did not control for poten-tial confounders in their analyses. For example, Pulver et al44found a
signifi-cantly (44-fold) increased risk of death for LPIs who were small for gestational age, compared with term infants with sizes appropriate for gestational age. Our findings revealed a slightly ele-vated aOR for neonatal death and a slightly higher aOR for postneonatal death; however, these results were not statistically significant. These differ-ences are likely explained by method-ologic differences between the afore-mentioned studies and the current study.
The current study excluded infants with birth defects, multiple births, and infants who were small for ges-tational age. Multivariate regression techniques were used to control for potential confounders, and we at-tempted to control for selection bias by using propensity score–matching methods. We think that our esti-mates of the mortality and morbidity rates for LPIs, compared with term infants, during the birth hospitaliza-tion more closely reflect the effects of the gestational age difference alone than do other studies con-ducted to date. Results from the cur-rent study suggest that LPIs are at greatest risk of death in the postneo-natal period, possibly because of in-creased risk of sudden unexpected infant death. Additional research is needed to confirm this suspicion. Un-like data from most other studies published to date, our data are reas-suring because they demonstrate only a nonsignificant trend toward increased mortality rates, although
many of the infants in this cohort were born in community nurseries without comprehensive neonato-logic support. Furthermore, al-though morbidity rates were in-creased in our LPI population, they were somewhat less than reported elsewhere, which reflects more-accurate estimates of gestational age effects alone.
This study includes several limita-tions that should be addressed. Data were obtained from administrative, claims, and birth certificate data-bases that were not necessarily signed for research. Much of the de-sign of the current study was based on the presence or absence of cer-tain diagnosis or procedure codes. Birth certificate data, hospital ad-ministrative data, and claims data are known to underreport many clin-ical conditions and procedures, to varying degrees.45–48 They also are
prone to clerical errors, misclassifi-cation, systematic trends in coding bias, and differential reimbursement rates. For example, if a diagnosis or procedure adds little to reimburse-ment, then it is much less likely to be coded than a highly reimbursed diag-nosis or procedure. For instance, ce-sarean section is coded at a consid-erably higher rate than episiotomy in hospital discharge data.49 However,
combining these data sources does improve accuracy over using the data sources alone.50,51Also, because
the data were not collected with this specific study in mind, omitted vari-able bias might be a significant con-cern, which cannot be addressed with the statistical and design meth-ods used in the current study. The data for the current study were lim-ited to the Medicaid population.
Hav-ing Medicaid as the primary payer source often is associated with lower socioeconomic status and gen-erally poorer outcomes, compared with having private insurance. How-ever, limiting the analysis to the Med-icaid population reduces somewhat the bias expected for populations with more-heterogeneous socioeco-nomic status.
CONCLUSIONS
Morbidity rates and health care utili-zation were increased for LPIs, com-pared with term neonates, even with the exclusion of neonates with birth defects and the use of propensity score–matching methods to control for obstetric factors. The risk of death was increased for LPIs, partic-ularly in the postneonatal period; however, this risk was not statisti-cally significant. Deliveries should not be scheduled during the late pre-term period without clear clinical in-dications. Clinicians should consider attempting to prolong otherwise-uncomplicated pregnancies that threaten labor in the late preterm period. Future research should focus on describing other areas of poten-tially increased morbidity among LPIs, including long-term morbidity, death after the neonatal period, and long-term health care utilization, and comparing these risks with the ben-efits of early delivery.
ACKNOWLEDGMENTS
REFERENCES
1. Davidoff MJ, Dias T, Damus K, et al. Changes in the gestational age distribution among U.S. singleton births: impact on rates of late preterm birth, 1992 to 2002.Semin Perina-tol.2006;30(1):8 –15
2. Engle WA, Tomashek KM, Wallman C. “Late-preterm” infants: a population at risk. Pedi-atrics.2007;120(6):1390 –1401
3. American Congress of Obstetricians and Gy-necologists, Committee on Obstetric Prac-tice. ACOG Committee Opinion No. 404, April 2008: late-preterm infants.Obstet Gynecol.
2008;111(4):1029 –1032
4. Fuchs K, Gyamfi C. The influence of obstetric practices on late prematurity.Clin Perina-tol.2008;35(2):343–360
5. Ecker JL, Frigoletto FD Jr. Cesarean delivery and the risk-benefit calculus.N Engl J Med.
2007;356(9):885– 888
6. Eichenwald EC, Stark AR. Management and outcomes of very low birth weight.N Engl J Med.2008;358(16):1700 –1711
7. Clapp DW. Developmental regulation of the immune system.Semin Perinatol. 2006; 30(2):69 –72
8. Schendel DE, Stockbauer JW, Hoffman HJ, Herman AA, Berg CJ, Schramm WF. Relation between very low birth weight and develop-mental delay among preschool children without disabilities.Am J Epidemiol.1997; 146(9):740 –749
9. Saigal S, Feeny D, Rosenbaum P, Furlong W, Burrows E, Stoskopf B. Self-perceived health status and health-related quality of life of extremely low-birth-weight infants at adolescence.JAMA.1996;276(6):453– 459 10. Gilbert WM, Nesbitt TS, Danielsen B. The cost
of prematurity: quantification by gesta-tional age and birth weight.Obstet Gynecol.
2003;102(3):488 – 492
11. Khashu M, Narayanan M, Bhargava S, Osiov-ich H. Perinatal outcomes associated with preterm birth at 33 to 36 weeks’ gestation: a population-based cohort study.Pediatrics.
2009;123(1):109 –113
12. Escobar GJ, Clark RH, Greene JD. Short-term outcomes of infants born at 35 and 36 weeks gestation: we need to ask more ques-tions.Semin Perinatol.2006;30(1):28 –33 13. Wang ML, Dorer DJ, Fleming MP, Catlin EA.
Clinical outcomes of near-term infants. Pe-diatrics.2004;114(2):372–376
14. Arnon S, Dolfin T, Litmanovitz I, Regev R, Bauer S, Fejgin M. Preterm labour at 34 –36 weeks of gestation: should it be arrested?
Paediatr Perinat Epidemiol. 2001;15(3): 252–256
15. Sarici SU, Serdar MA, Korkmaz A, et al.
Inci-dence, course, and prediction of hyperbil-irubinemia in near-term and term new-borns.Pediatrics.2004;113(4):775–780 16. Seubert DE, Stetzer BP, Wolfe HM, Treadwell
MC. Delivery of the marginally preterm infant: what are the minor morbidities?
A m J O b s t e t G y n e c o l . 1 9 9 9 ; 1 8 1 ( 5 ) : 1087–1091
17. Bhutani VK, Johnson L. Kernicterus in late preterm infants cared for as term healthy infants.Semin Perinatol.2006;30(2):89 –97 18. Melamed N, Klinger G, Tenenbaum-Gavish K,
et al. Short-term neonatal outcome in low-risk, spontaneous, singleton, late preterm deliveries. Obstet Gynecol. 2009;114(2): 253–260
19. Escobar GJ, Greene JD, Hulac P, et al. Rehos-pitalisation after birth hosRehos-pitalisation: pat-terns among infants of all gestations.Arch Dis Child.2005;90(2):125–131
20. Shapiro-Mendoza CK, Tomashek KM, Kotel-chuck M, Barfield W, Weiss J, Evans S. Risk factors for neonatal morbidity and mortal-ity among “healthy,” late preterm new-borns.Semin Perinatol.2006;30(2):54 – 60 21. Tomashek KM, Shapiro-Mendoza CK, Weiss
J, et al. Early discharge among late preterm and term newborns and risk of neonatal morbidity. Semin Perinatol. 2006;30(2): 61– 68
22. Sibai BM. Preeclampsia as a cause of pre-term and late prepre-term (near-pre-term) births.
Semin Perinatol.2006;30(1):16 –19 23. Laptook A, Jackson GL. Cold stress and
hy-poglycemia in the late preterm (“near-term”) infant: impact on nursery of admis-sion.Semin Perinatol.2006;30(1):24 –27 24. Blair E, Watson L. Epidemiology of cerebral
palsy. Semin Fetal Neonatal Med.2006; 11(2):117–125
25. Jain L, Eaton DC. Physiology of fetal lung fluid clearance and the effect of labor. Se-min Perinatol.2006;30(1):34 – 43
26. Hansen AK, Wisborg K, Uldbjerg N, Henrik-sen TB. Elective caesarean section and re-spiratory morbidity in the term and near-term neonate.Acta Obstet Gynecol Scand.
2007;86(4):389 –394
27. Kramer MS, Demissie K, Yang H, Platt RW, Sauve R, Liston R. The contribution of mild and moderate preterm birth to infant mor-tality.JAMA.2000;284(7):843– 849 28. Tomashek KM, Shapiro-Mendoza CK,
David-off MJ, Petrini JR. Differences in mortality between late-preterm and term singleton infants in the United States, 1995–2002.
J Pediatr.2007;151(5):450 – 456
29. McLaurin KK, Hall CB, Jackson EA, Owens OV,
Mahadevia PJ. Persistence of morbidity and cost differences between late-preterm and term infants during the first year of life. Pe-diatrics.2009;123(2):653– 659
30. Huddy CL, Johnson A, Hope PL. Educational and behavioural problems in babies of 32–35 weeks gestation.Arch Dis Child Fetal Neonatal Ed.2001;85(1):F23–F28
31. Boyce TG, Mellen BG, Mitchel EF Jr, Wright PF, Griffin MR. Rates of hospitalization for respiratory syncytial virus infection among children in Medicaid.J Pediatr.2000;137(6): 865– 870
32. Klassen AF, Lee SK, Raina P, Chan HW, Mat-thew D, Brabyn D. Health status and health-related quality of life in a population-based sample of neonatal intensive care unit grad-uates.Pediatrics.2004;113(3):594 – 600 33. 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(1):25–31
34. Petrini JR, Dias T, McCormick MC, Massolo ML, Green NS, Escobar GJ. Increased risk of adverse neurological development for late preterm infants.J Pediatr. 2009;154(2): 169 –176
35. Rosenbaum PR, Rubin DB. The central role of the propensity score in observational studies for causal effects.Biometrika.1983; 70(1):41–55
36. Rubin DB. Estimating causal effects from large data sets using propensity scores.
Ann Intern Med.1997;127(8):757–763 37. Bronstein JM, Lomatsch CT, Fletcher D, et al.
Issues and biases in matching Medicaid pregnancy episodes to vital records data: the Arkansas experience. Matern Child Health J.2009;13(2):250 –259
38. Bird TM, Hobbs CA, Cleves MA, Tilford JM, Robbins JM. National rates of birth defects among hospitalized newborns.Birth De-fects Res A Clin Mol Teratol.2006;76(11): 762–769
39. Robbins J, Bird T, Tilford J, et al. Hospital stays, hospital charges, and in-hospital deaths among infants with selected birth defects: United States, 2003.MMWR Morb Mortal Wkly Rep.2007;56(2):25–29 40. Silber JH, Rosenbaum PR, Trudeau ME, et al.
Multivariate matching and bias reduction in the surgical outcomes study. Med Care.
2001;39(10):1048 –1064
Medicare. Health Serv Res. 2004;39(6): 1773–1792
42. Basu A, Manning WG. Issues for the next gen-eration of health care cost analyses.Med Care.2009;47(7 suppl 1):S109 –S114 43. Austin PC, Ghali WA, Tu JV. A comparison of
several regression models for analysing cost of CABG surgery.Stat Med.2003;22(17): 2799 –2815
44. Pulver LS, Guest-Warnick G, Stoddard GJ, By-ington CL, Young PC. Weight for gestational age affects the mortality of late preterm in-fants.Pediatrics.2009;123(6). Available at: www. pediatrics.org/cgi/content/full/123/6/e1072
45. DiGiuseppe DL, Aron DC, Ranbom L, Harper DL, Rosenthal GE. Reliability of birth
certifi-cate data: a multi-hospital comparison to medical records information.Matern Child Health J.2002;6(3):169 –179
46. Roohan PJ, Josberger RE, Acar J, Dabir P, Feder HM, Gagliano PJ. Validation of birth certificate data in New York State.J Com-munity Health.2003;28(5):335–346 47. Korst LM, Gregory KD, Gornbein JA. Elective
primary caesarean delivery: accuracy of ad-ministrative data.Paediatr Perinat Epide-miol.2004;18(2):112–119
48. Yasmeen S, Romano PS, Schembri ME, Keyzer JM, Gilbert WM. Accuracy of obstet-ric diagnoses and procedures in hospital discharge data.Am J Obstet Gynecol.2006; 194(4):992–1001
49. Chung A, Macario A, El-Sayed YY, Riley ET, Duncan B, Druzin ML. Cost-effectiveness of a
trial of labor after previous cesarean. Ob-stet Gynecol.2001;97(6):932–941
50. Lydon-Rochelle MT, Holt VL, Nelson JC, et al. Accuracy of reporting maternal in-hospital diagnoses and intrapartum procedures in Washington State linked birth records. Pae-diatr Perinat Epidemiol. 2005;19(6): 460 – 471
DOI: 10.1542/peds.2009-2869 originally published online July 5, 2010;
2010;126;e311
Pediatrics
and Glen P. Mays
T. Mac Bird, Janet M. Bronstein, Richard W. Hall, Curtis L. Lowery, Richard Nugent
Year
Late Preterm Infants: Birth Outcomes and Health Care Utilization in the First
Services
Updated Information &
http://pediatrics.aappublications.org/content/126/2/e311
including high resolution figures, can be found at:
References
http://pediatrics.aappublications.org/content/126/2/e311#BIBL
This article cites 50 articles, 8 of which you can access for free at:
Subspecialty Collections
http://www.aappublications.org/cgi/collection/neonatology_sub
Neonatology
sub
http://www.aappublications.org/cgi/collection/fetus:newborn_infant_
Fetus/Newborn Infant following collection(s):
This article, along with others on similar topics, appears in the
Permissions & Licensing
http://www.aappublications.org/site/misc/Permissions.xhtml
in its entirety can be found online at:
Information about reproducing this article in parts (figures, tables) or
Reprints
http://www.aappublications.org/site/misc/reprints.xhtml
DOI: 10.1542/peds.2009-2869 originally published online July 5, 2010;
2010;126;e311
Pediatrics
and Glen P. Mays
T. Mac Bird, Janet M. Bronstein, Richard W. Hall, Curtis L. Lowery, Richard Nugent
http://pediatrics.aappublications.org/content/126/2/e311
located on the World Wide Web at:
The online version of this article, along with updated information and services, is
by the American Academy of Pediatrics. All rights reserved. Print ISSN: 1073-0397.