Copyright 2001 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2001
To determine the risk of hospitalization associated with prenatal care use and indicators of socioeconomic status and social support among infants of teenage mothers.
Population-based case-control study.
Nonfederal hospitals in Washington State.
Infants born from 1987 to 1995 to mothers younger than 20 years were identified using linked birth certificate–state hospital discharge data. Cases consisted of 8052 infants who were hospitalized during the first year of life at least 2 days after birth hospitalization discharge. An equal number of controls, frequency matched on birth year and maternal age group, were randomly sampled from among nonhospitalized infants.
Main Outcome Measure
Hospitalization in the first year of life.
Infants with a father listed on the birth certificate or whose mothers had commercial health insurance had a decreased risk of hospitalization (adjusted odds ratios, 0.91 and 0.78, respectively; 95% confidence intervals, 0.83-0.99 and 0.71-0.85, respectively). Participation in state-funded pregnancy programs, adequacy of prenatal care, or marital status did not affect the risk of hospitalization, except among infants whose mothers received more than adequate prenatal care (adjusted odds ratio, 1.15; 95% confidence interval, 1.03-1.29).
Our results suggest that teenaged mothers who list a father on the birth certificate or who have insurance, indicative of higher socioeconomic status, may have a reduced risk of hospitalization for their infants. Teenaged mothers who receive more than adequate prenatal care may have pregnancy complications that place their infants at increased risk of hospitalization. The effect of these protective factors should be clarified in future studies.
TEEN PREGNANCY represents a major public health problem, with more than 500 000 births annually to teenage mothers in the United States.1,2 Teen pregnancy is associated with many adverse neonatal outcomes, including prematurity, low birth weight, and neonatal death.3- 6 It is also recognized that infants born to teenage mothers are at increased risk of hospitalization during the first year of life, relative to infants of older mothers.7- 10 The majority of these hospital admissions are for potentially avoidable causes such as gastroenteritis, respiratory infections, and injuries.8- 10
Identification of factors associated with hospitalizations among infants of teenaged mothers may aid the design of interventions to reduce this risk. While previous studies have identified many risk factors associated with hospitalization, such as lower socioeconomic status, maternal illness during pregnancy, presence of congenital anomalies, low birth weight, male sex, and the presence of other siblings in the family,7- 10 these studies have been unable to reliably identify all infants at greater risk.10 Consequently, it has been recommended that interventions be targeted to all teenaged mothers. However, it is unclear what components should be included in such interventions.
We undertook a population-based, case-control study to examine the association of prenatal care use and indicators of socioeconomic status and social support with hospitalization in the first year of life among infants born to teenaged mothers. We hypothesized that factors indicative of higher socioeconomic status, social support during pregnancy, and adequacy of prenatal care would be associated with a decreased risk of subsequent hospitalization for these infants.
Data were obtained from the Washington State Department of Health (DOH), which compiles information on all births and nonfederal hospital discharges in Washington State on separate databases. The Comprehensive Hospital Abstract Reporting System (CHARS) is a state-administered database that contains information on all nonfederal hospital discharges in Washington State, beginning in 1987. Hospital discharge data from these hospitals are reported to the DOH and are edited for internal and logical consistency. The data are not routinely evaluated for validity. The CHARS reports data on reasons for hospitalization (up to 9, using International Classification of Disease, Ninth Revision[ICD-9] codes), length of stay, dates of admission and discharge, primary payer, and procedure codes. The Birth Events Records Database (BERD) is a state-administered database that links birth certificate data for all infants born in nonfederal hospitals with hospital discharge data related to the birth hospitalizations of mothers and their newborns. The BERD reports information on date of birth; infant's, mother's, and father's demographic characteristics; prenatal visits; prior pregnancies and births; pregnancy and birth history; pregnancy program participation; and the agency or type of payer to which the hospital charges were billed. Linkage algorithms used by the DOH to create BERD have demonstrated a greater than 98% correct match rate (personal communication, Vicki Hohner, MBA, Office of Hospital and Patient Data Systems, DOH, November 24, 1999). Birth certificate data undergo quality checks for completeness, validity, and internal consistency before they are included in BERD. These checks are performed at the time of data entry, at uploads of records to the main database, and at regular reviews of the main database (weekly, quarterly, and annually) by DOH, other state, and federal employees.
All singleton infants born in Washington State from 1987 to 1995 to mothers younger than 20 years were identified using BERD. Cases consisted of all infants who were hospitalized in Washington State during the first 12 months of life. Cases were identified by linking the personal identifier code in the BERD file for the years 1987 to 1995 with that from the CHARS file for the years 1987 to 1996. Only first hospitalizations were considered for analysis. Cases were excluded if the hospitalization occurred within the first 2 days after birth hospitalization, to avoid inclusion of hospitalizations for possible birth complications. One control per case was selected at random from among infants who were not hospitalized in the first year of life. Controls were frequency matched to cases on maternal age group (≤17 years, 18-19 years) and year of birth. The study protocol was approved by the Human Subjects Protection Committees at the University of Washington and the DOH.
The main outcome measure was hospitalization in the first year of life for any reason. Hospitalizations among cases were classified by specific causes: respiratory syncytial virus (RSV) lower respiratory tract disease (ICD-9 codes 466.1, 480.1); neonatal jaundice (ICD-9 code 774.6); non-RSV pneumonia (ICD-9 codes 480-486, except 480.1); gastroenteritis (ICD-9 codes 008, 009, and 558.9); viral infections (ICD-9 codes 045-079); dehydration (ICD-9 code 276.5); fever (ICD-9 code 780.6); injuries and poisonings (E-codes 800-999); and other.
Associations between hospitalization and prenatal care or selected factors related to socioeconomic status and social support were evaluated. The type of health insurance billed as the primary payer for the mother's birth hospitalization was used as a proxy indicator of socioeconomic status. Medicaid/charity care represented lower socioeconomic status and health maintenance organization, commercial, or other insurance represented higher socioeconomic status. Medicaid managed care was classified as Medicaid. Adequacy of prenatal care was measured using the Adequacy of Prenatal Care Utilization Index of Kotelchuck.11,12 The Adequancy of Prenatal Care Utilization Index categorizes prenatal care into 4 groups based on the timing of initiation of prenatal care and the observed-to-expected number of prenatal care visits: inadequate (initiation of prenatal care after the fourth month of gestation or less than 50% of expected prenatal care visits); intermediate (initiation of prenatal care before the fourth month of gestation and between 50% and 80% of expected prenatal care visits); adequate (initiation of prenatal care before the fourth month of gestation and between 80% and 110% of expected visits); and adequate plus (initiation of prenatal care before the fourth month of gestation and greater than 110% of expected prenatal care visits). Three variables were used as indicators of social support: marital status, whether a father was listed on the birth certificate, and whether the mother participated in state-administered pregnancy programs. To list a father on a birth certificate in Washington State during the study period, the parents must have been married or filed an affidavit of paternity. An affidavit of paternity ensures that listed fathers are legally and financially responsible for the care of their infants. In the few cases where paternity was in dispute, a father could be listed after results of paternity testing. The following state-administered programs were available to eligible pregnant women in Washington State during the study period: Women, Infants, and Children; Aid to Families With Dependent Children; First Steps, the Washington State Medicaid expansion program; and other programs.
Other variables were considered in the analysis for their possible effect on the variables of interest. These variables were derived from birth certificate data and included birth year; infant sex; birth weight (<2500 g, 2500-4000 g, or >4000 g); estimated gestational age (<37 weeks or ≥37 weeks); any congenital anomaly; any newborn medical condition; maternal age group (<18 years or 18-19 years); maternal self-report of smoking; maternal self-report of drinking during pregnancy (not routinely collected on birth certificates until 1989); maternal race/ethnicity (white, black, Native American, Asian, or Hispanic); urban or rural maternal residence; maternal gravidity and parity (none, ≥1); history of prior induced pregnancy terminations (none, ≥1); and history of any prior child deaths.
Univariate associations between hospitalization and the primary variables of interest were evaluated by the odds ratio (OR) with 95% confidence intervals (CIs). Stratified analyses were performed using Mantel-Haenszel estimators, with test-based CIs to control for potential confounding variables and assess for interactions.13 The Breslow-Day Test was used to evaluate the homogeneity of stratum-specific ORs.14 Only variables that markedly altered the risk estimates or resulted in significant interactions were retained in the analyses. Logistic regression was used to estimate the multivariate association between hospitalization and the primary variables of interest while controlling for any other variables retained in the analysis. Statistical analyses were performed using STATA, Release 5.0 for Macintosh (College Station, Tex).
A total of 74 382 singleton births to teenaged mothers were identified from Washington State birth records during the study period 1987 to 1995. Of these, 8492 infants of teenaged mothers were hospitalized within the first year of life. We excluded 440 infants (5.2%) who were admitted within 2 days of birth discharge to avoid hospitalizations for possible birth complications. Thus, 8052 infants (10.8%) were hospitalized at least 2 days after hospital discharge and constituted cases for analysis. These infants were hospitalized for a variety of reasons, most commonly for RSV lower respiratory tract disease (17.4%), jaundice (8.6%), non-RSV pneumonia (8.5%), viral syndrome (5.8%), gastroenteritis (4.5%), and injuries and poisonings (3.8%). Most infants (84.2%) incurred only 1 hospitalization. However, 927 (11.5%) had 2 hospitalizations and 342 (4.3%) had 3 or more hospitalizations in the first year of life.
During the study period, rates of hospitalization for infants by birth year remained fairly stable. Rates ranged from 102.4 per 1000 in 1987 to 116.8 per 1000 in 1991. Rates of hospitalization were similar for infants of mothers 17 years or younger and aged 18 to 19 years. During the study period, the percentage of singleton births to teenaged mothers in Washington State remained relatively constant (approximately 11%).
Most teenaged mothers in Washington State during the study period were white, resided in urban areas, and had no prior pregnancies or live births. Although most teenaged mothers (62.8%) were aged 18 to 19 years, the majority of fathers (68.8%) were 20 years and older. There were 1111 (6.9%) teenaged mothers who were younger than 16 years.
Mothers of cases and controls were similar with respect to paternal age, maternal race other than Native American, previous induced abortions, previous child deaths, and self-reported use of alcohol during pregnancy (Table 1). A greater percentage of mothers of cases were Native American (7.1% vs 4.4%), resided in urban locations (64.1% vs 60.9%), had a previous pregnancy (37.3% vs 30.7%) or live birth (25.1% vs 19.8%), and reported smoking during pregnancy (36.9% vs 31.1%) relative to mothers of controls.
Most infants were born full-term, weight appropriate for gestational age, and without congenital anomalies or newborn medical conditions (Table 2). A greater percentage of cases were of low birth weight (10.9% vs 5.6%), premature (13.0% vs 7.1%), male (58.4% vs 48.7%), diagnosed with a medical condition at birth (14.0% vs 11.3%), or born with a congenital anomaly (5.0% vs 2.7%) relative to controls.
Most teenaged mothers (58.8%) began prenatal care within the first 3 months of gestation. A greater percentage of mothers of cases received more than adequate prenatal care (25.4% vs 21.6%) relative to mothers of controls (Table 3). Infants whose mothers received more than adequate prenatal care had an increased risk of hospitalization (OR, 1.15; 95% CI, 1.03-1.29; adjusted for gestational age, socioeconomic status, marital status, listing a father, and pregnancy program participation) relative to infants whose mothers received inadequate prenatal care. Infants whose mothers received intermediate or adequate prenatal care did not have a risk of hospitalization different from infants whose mothers received inadequate prenatal care.
Regarding indicators of social support (Table 3), approximately one quarter of teenaged mothers in the study were married (28%). Slightly fewer mothers of cases than controls were married (27.0% vs 29.1%). A majority of infants had a father listed on the birth certificate (53.2%). A slight majority of mothers who listed a father on the birth certificate were married (51.5%). A smaller percentage of cases had a father listed on the birth certificate (51.6% vs 54.7%) relative to controls. One third of teenaged mothers (34.4%) participated in state-administered pregnancy programs. Among program users, most participated in Women, Infants, and Children (84.9%), followed by First Steps (40.5%), Aid to Families With Dependent Children (19.2%), and other miscellaneous programs (19.1%). A slightly greater percentage of mothers of cases than controls participated in these programs (35.3% vs 33.6%). Infants with a father listed on the birth certificate had a decreased risk of hospitalization relative to infants without a father listed on the birth certificate (OR, 0.91; 95% CI, 0.83-0.99, adjusted for gestational age, prenatal care, socioeconomic status, marital status, and pregnancy program participation). Infants whose mothers were married or whose mothers participated in pregnancy programs did not have a risk of hospitalization different from those whose mothers were single or whose mothers did not participate in pregnancy programs.
Most teenaged mothers (72.3%) had Medicaid insurance or charity care listed as the primary payer at the time of birth hospitalization. A greater percentage of mothers of cases than controls had Medicaid insurance or received charity care (Table 3). Mothers of cases had significantly lower mean years of education (10.3 vs 10.5 years) than mothers of controls, but the difference was clinically insignificant. Data on maternal education were not routinely reported on birth certificates in the state until 1992. Infants whose mothers had commercial, health maintenance organization, or other health insurance listed as the payer on the birth certificate had a decreased risk of hospitalization (OR, 0.78; 95% CI, 0.71-0.85; adjusted for gestational age, prenatal care, marital status, listing a father, and pregnancy program participation) relative to infants whose mothers had Medicaid insurance or charity care listed as the payer.
Teenage mothers who were younger than 16 years at the time of birth hospitalization were analyzed separately. There was no significant difference between cases and controls in the percentage of teenage mothers younger than 16 years (6.8% of cases vs 7.0% of controls). Teenage mothers younger than 16 years were more likely to have inadequate prenatal care (38% vs 27%) and use pregnancy programs (39.5% vs 34.1%) than older mothers. Moreover, teenage mothers younger than 16 years were less likely to be married (6.1% vs 29.6%) or to list a father on the birth certificate (31.2% vs 54.8%) than older mothers. Infants of teenaged mothers who were younger than 16 years had risk estimates that were similar to those obtained from the entire sample, except that the point estimate obtained for the risk of hospitalization associated with being married increased (adjusted OR, 1.63), and the point estimate obtained for the risk of hospitalization associated with more than adequate prenatal care decreased (adjusted OR, 1.03). None of the point estimates from this subgroup analysis was significant.
A subgroup analysis was undertaken to evaluate the effect of infant factors known or suspected to increase the risk of hospitalization (low birth weight, prematurity, congenital anomalies, and neonatal conditions) on adjusted risk estimates. To assess whether these findings were consistent among infants without these characteristics, the analysis was repeated after restricting infants to those who were of normal birth weight (2500-4000 g), term gestation (>37 weeks), and without congenital anomalies or newborn medical conditions. A total of 7923 infants (49.2%) with at least 1 risk factor were omitted from this subgroup analysis. The resulting risk estimates did not differ markedly from those obtained with the entire sample.
A second subgroup analysis was undertaken to evaluate whether hospitalizations for possible birth complications influenced adjusted risk estimates. To evaluate for this, the analysis was repeated using only cases whose hospital admission occurred after 28 days of life. A total of 2313 infants (28.7%) who were hospitalized within the first 28 days of life were omitted from this subgroup analysis. There were no appreciable changes in the risk estimates from those obtained with the entire sample.
The risks of hospitalization for specific conditions were evaluated. The specific conditions included RSV lower respiratory tract disease, jaundice, non-RSV pneumonia, and injuries and poisonings (E-codes). The risks of hospitalization for RSV lower respiratory tract disease (N = 1398) were similar to the overall risks of hospitalization, except that the risk associated with receiving more than adequate prenatal care increased (adjusted OR, 1.28; 95% CI, 1.05-1.57). The risks of hospitalization for jaundice (n = 694) were similar, except that the risk associated with receiving more than adequate prenatal care increased (adjusted OR, 1.85; 95% CI, 1.42-2.42) and the risk of being married became significant (adjusted OR, 1.36; 95% CI, 1.08-1.72). The risks of hospitalization for injuries and poisonings (n = 306) or non-RSV pneumonia (n = 685) were similar to the overall risks of hospitalization in the entire sample.
We performed, to our knowledge, the first population-based analysis of infants of teenaged mothers and the risk of hospitalization in the first year of life associated with possible protective factors. We found that infants whose mothers have commercial insurance as the primary payer had a 22% reduction in the risk of hospitalization. We also found that infants whose fathers were listed on the birth certificate had a 9% reduction in the risk of hospitalization. This latter risk reduction represents more than $3.1 million in averted hospital charges in Washington State during the study period based on mean hospital charges. We found that marital status, participation in state-administered pregnancy programs, and adequacy of prenatal care did not affect the risk of infant hospitalization except among women receiving the highest category of prenatal care. Here, infants whose mothers received more than adequate prenatal care had a 15% increased risk of hospitalization. This increased risk occurred primarily among those infants hospitalized for RSV lower respiratory tract disease and newborn jaundice. Risks of hospitalization did not vary appreciably among healthy term infants of normal birth weight or those hospitalized at older than 28 days. Risks of hospitalization varied slightly between mothers younger than 16 years and those who were older.
Findings from the literature suggest that teenaged mothers who receive social support may reduce their risk of many adverse pregnancy outcomes, such as low birth weight.15- 17 However, the few studies to date reporting on hospitalization among infants of teenaged mothers have not examined the effect of paternal social support.7- 10 Studies reporting on other outcomes have found that teenaged mothers whose partners participate in the care of their children have more positive child-rearing attitudes18,19 and less depressive symptoms.20 In our study, we found that infants who had a father listed on the birth certificate had a decreased risk of hospitalization while infants whose mothers were married or participated in pregnancy programs did not have risk different from infants whose mothers were not. In most cases, teenaged parents must be legally married or file an affidavit of paternity to list a father on the birth certificate in Washington State. Fathers who complete an affidavit of paternity become legally and financially responsible for the care of their children. We speculate that teenaged mothers in this study who listed a father on the birth certificate may have been involved in more stable relationships that support the care of their children. However, we did not specifically measure paternal social support, and this finding must be weighed against the finding that marital status was not associated with risk of hospitalization.
The finding of a decreased risk of hospitalization associated with an indicator of higher socioeconomic status, use of commercial insurance, is consistent with a large body of evidence supporting the association of lower socioeconomic status and adverse pregnancy outcomes.6,8,21,22 The reasons for such a finding are complex and are likely to be related to many factors associated with lower socioeconomic status, such as poor access to primary care, transportation difficulties, lack of social supports, cultural barriers, and high-risk behaviors.23
Although other studies have shown that adequate prenatal care can reduce the risk of adverse pregnancy outcomes,12,16,24 we did not find a significant association between the level of prenatal care and later risk of infant hospitalization, except for those who received more than adequate prenatal care. Adequate prenatal care reflects both sufficient access to care and knowledge and motivation on the part of women to appropriately use care. It may be that prenatal care is not a good proxy measure of health care access after delivery because of changes in insurance coverage. However, more than adequate prenatal care may identify teenaged mothers who have pregnancy-related complications that place their child at greater risk of hospitalization in the first year of life.12
Our study has several important strengths. First, our study was population based and identified all singleton infants of teenaged mothers in Washington State; previous studies have not been population based. Second, we used a comprehensive statewide hospital discharge database that enabled us to identify virtually all cases of infants of teenaged mothers who were hospitalized. Only those few infants whose mothers moved out of state or went across state lines to have their infants hospitalized would have been missed. Previous studies employing survey techniques to classify cases may be limited by inaccuracies in maternal recall of hospitalizations in the first year of life.25 Third, we relied on birth certificate information gathered at the time of delivery to categorize variables of interest. Previous studies using survey techniques may be limited by recall bias, in which cases they may differentially recall past events relative to controls. Fourth, we were able to identify a large number of cases that enabled us to have sufficient power to detect small associations. Previous studies have been smaller and may not have had sufficient power to detect the small associations that we observed in this study.
Our study does have several limitations. First, the use of administrative databases in general may contain errors in diagnostic coding. However, there is no reason to suspect that differential misclassification of information occurred preferentially in either cases or controls. Any bias in such an instance would be towards the null. Second, information on health insurance at the time of rehospitalization was not available for controls, so we used the primary payer for the birth hospitalization as an indicator of socioeconomic status. Since insurance status may change from the time of birth hospitalization to rehospitalization, any such changes may alter the risk of hospitalization associated with this variable. For cases, we found that 86.8% of those originally classified with Medicaid insurance or receiving charity care at the time of birth hospitalization were similarly classified at the time of rehospitalization, suggesting that any changes in health insurance in the study population were probably small. Third, information on the health status of infants at the time of hospitalization was not available. However, we did restrict the analysis to those infants who were without prematurity, low birth weight, congenital anomalies, and newborn medical conditions, which are the main determinants of poor infant health status in the first year of life. Results of the subgroup analysis were not different from those results obtained from the entire population. Fourth, identification of infants hospitalized within the first month of life may include hospitalizations for birth complications. Restriction of cases to those infants hospitalized at older than 28 days did not alter risk estimates appreciably. Fifth, our use of insurance type as a proxy indicator of socioeconomic status may have limited utility. However, we found that subjects with Medicaid or charity care had a significantly lower mean income ($22 450 vs $24 751; P<.001) than subjects with health maintenance organization, commercial, or other insurance types. Moreover, Medicaid managed care was not initiated in the state until 1993, so its effect was not apparent during most of the study period. Coding algorithms from the DOH specified that Medicaid managed care was to be coded as Medicaid on all payer fields. Sixth, information on hospitalizations from federal hospitals in the state was not included in the analysis. Births and hospitalizations that occur in federal hospitals (eg, military facilities) represent a small percentage of the total. It is unclear what effect the absence of these hospitalizations has on overall risk estimates.
This study has possible implications for the design of interventions to reduce the risk of hospitalization among infants born to teenaged mothers. Although the incorporation of social support interventions before delivery may reduce adverse pregnancy outcomes, it is unclear whether such interventions after delivery may affect adverse infant outcomes such as hospitalization. Our data suggest a reduced risk of infant hospitalization among teenaged mothers who list a father on the birth certificate. Further studies of infant hospitalization that incorporate more explicit measures of social support from partners and family members may help to clarify these findings, particularly in the context of marital status.
Accepted for publication September 6, 2000.
Presented in part at the regional meeting of the Ambulatory Pediatrics Association, Carmel, Calif, January 31, 1999.
We thank the Washington State Department of Health, for providing access to these data and William O'Brien, BS, for his assistance in the programming of the dataset.
Corresponding author: James Guevara, MD, MPH, Division of General Pediatrics, Children's Hospital of Philadelphia, 34th and Civic Center Boulevard, Philadelphia, PA 19104.
Guevara JP, Young JCC, Mueller BA. Do Protective Factors Reduce the Risk of Hospitalization in Infants of Teenaged Mothers?. Arch Pediatr Adolesc Med. 2001;155(1):66-72. doi:10.1001/archpedi.155.1.66