Copyright 2001 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2001
To investigate behavioral risks and life circumstances of adolescent mothers with older (≥5 years) adult (≥20 years old) vs similar-aged (±2 years) male partners at 12 months' postpartum.
Nine hundred thirty-one adolescent females were interviewed after delivery and were mailed surveys to complete at 12 months' postpartum. Analysis by χ2 and t test was used to identify differences in behavioral risks (planned repeated pregnancy, substance use, and intimate partner violence) and life circumstances (financial status, school enrollment, and social support) for adolescent mothers with older adult vs similar-aged partners. Additional stratified analyses were conducted to evaluate the extent to which living with an adult authority figure or being with the father of her infant born 12 months previously might alter observed relationships.
At 12 months following delivery, 184 adolescent mothers (20%) reported having an older adult partner, whereas 312 (34%) had a similar-aged partner. The remaining adolescent mothers (n = 239) were excluded from further analyses. Adolescent mothers with older adult partners were significantly less likely to be employed or enrolled in school and were more likely to report planned repeated pregnancies. These adolescent mothers also received less social support. No differences were observed in intimate partner violence or the mother's substance use. Adolescent mothers with older adult partners who did not live with an adult authority figure seemed to be at greatest risk.
The negative educational and financial impact of coupling with an older vs similar-aged partner seems greater for those mothers who no longer reside with an adult authority figure. These adolescent mothers are also at greater risk of planned rapid repeated pregnancy. Given their limited educational attainment and family support, a subsequent pregnancy may place these young women at considerable financial and educational disadvantage.
ADULT MALES father more than 50% of all live births to adolescent girls; of these, more than 40% involve fathers who are at least 5 years older than their partners.1- 3
Age-discrepant couples tend to have great differences in maturity, sexual and life experiences, social position, financial resources, and educational attainment. There has been increasing interest in enforcing previously widely ignored statutory rape laws to reduce rates of teenage pregnancy and protect vulnerable young women.4,5 These recent notable events have promulgated research into evaluating the effects of age-discrepant pairings.
Some investigators have reported that older adult male partners of pregnant or parenting adolescent females have a lifestyle that includes frequent tobacco, alcohol, and other drug use.6,7
Others have found that adult fathers of infants born to adolescent mothers are no more likely to engage in harmful substance use than adolescent fathers.8 Violence also has been implied in these age-discrepant relationships.4 However, a recent study of pregnant adolescents found that intimate partner violence toward adolescent mothers—physical or sexual assault—was no more frequent among older adult partners than among similar-aged partners.6
In addition, several findings support the popular lay opinion that older adult partners may encourage pregnancy at higher rates than similar-aged partners. For example, among women younger than 18 years, those with older partners are more likely to become pregnant and to report that the pregnancy was planned than those whose partners were no more than 2 years older.9
They are also less likely to report using birth control or condoms at last intercourse.9 Intimate partner violence among age-discrepant couples following the birth of a child and more subtle forms of coercion and control, such as pressuring an adolescent mother to become pregnant again or promoting her dependence by isolating her from family and peers, have not been systematically examined.
Adolescent females who become pregnant by older men are more likely to be living with their partners,10 and adolescent mothers who live with their partners are less likely to be enrolled in school.11 In contrast, adolescent mothers who continue to live with family members following delivery are more likely to return to school, graduate, and become employed and independent of welfare assistance.12 Thus, continued residence with family members may be an important mediating variable between the effects of coupling with an older adult male and outcomes to adolescent mothers. For example, adolescent mothers involved with older partners who do not live with parents or other adult figures may be more vulnerable to sexual coercion, intimate partner violence, or the use of harmful substances. In addition, these young mothers may be less likely to be enrolled in school because they do not have supportive networks for so doing. Yet, the mediating role of continued residence with family members on age-discrepant pairings and outcomes to adolescent mothers has not been examined.
Older adult partners may also provide unique benefits for young mothers. They seem to be more interested in and capable of assuming child-rearing responsibilities,3,7 and children reared in father-present vs father-absent homes have been found to have better outcomes.13,14
These men are more likely to be employed and may therefore have greater immediate earning potential than similar-aged partners.4
In addition, adolescent females report that they are often treated better by older partners compared with adolescent partners.4
Finally, some cultures view the relationship between adolescent females and older adult partners as desirable and actively encourage such couplings.4
To clarify and expand on the limited research already conducted on pregnant adolescents, systematic inquiry into the lives of young women with older adult partners into the postpartum period is necessary. The purpose of this study was to compare behavioral risks (planned repeated pregnancy, substance use, violence, and contraception) and life circumstances (financial status, school enrollment, social support, and isolation) of 2 groups of adolescent mothers: those with older (≥ 5 years) adult (≥20 years old) partners and those coupled with similar-aged (±2 years) male partners 1 year following delivery. A second purpose was to evaluate whether continuing to live with an adult authority figure (parents, grandparents, aunt/uncle, or other guardian) mediates these potential risks and life circumstances. A final purpose was to assess whether these risks and life circumstances were different for adolescent mothers with older adult partners vs similar-aged partners who were and were not the fathers of their infants born 12 months previously. It was hypothesized that adolescent mothers with older adult partners as compared with similar-aged partners would be more likely to report substance use by partner or self, intimate partner violence toward the adolescent mother, social isolation, and planned repeated pregnancy. It was further hypothesized that these relationships would be stronger among adolescent mothers not living with an adult authority figure and those whose current partner was not the father of their previous child.
All adolescent mothers through 18 years of age who delivered at The University of Texas Medical Branch at Galveston between December 8, 1993, and February 28, 1996, were eligible to participate if they met the following 5 criteria: (1) self-identified race of Mexican American, African American, or white; (2) planned to retain custody of her infant; (3) ability to read and write at a fifth-grade level in either English or Spanish; (4) absence of major psychiatric disorders; and (5) delivery of a healthy infant weighing more than 1500 g. Adolescent mothers from other racial/ethnic groups were excluded from study participation because there were few in each group. There were 1053 adolescent mothers who delivered their infants during the study period and were eligible to participate. Of these, 26 were not approached because of the numerous births on the days they gave birth. Of the 1027 adolescent mothers invited to participate in the study, 96 refused; the most common reason given was insufficient time to complete the hour-long interview. The study sample therefore included 931 mothers: 281 white, 349 Mexican American, and 301 African American. Demographic comparisons between those who refused to participate and those who were interviewed revealed a higher refusal rate among Mexican American mothers who spoke only Spanish (23%, P<.001). Written consent in English or Spanish was obtained from each participant as well as from a parent or legal guardian for adolescent mothers younger than 18 years who lived at home and were not married.
With institutional review board approval, a trained female research assistant interviewed each study subject privately in English or Spanish in the postpartum ward within 48 hours of delivery. Interviewers who spoke fluent Spanish interviewed all patients who preferred to converse in Spanish. Data were collected as part of a larger study of the transition to adolescent parenthood, which focused in large part on substance use.15
Demographic information was obtained from each adolescent mother during the face-to-face structured interview. As part of the larger study, each agreed to complete a self-report survey at 12 months' postpartum. These follow-up surveys were mailed; telephone interviews were carried out if requested. To maximize survey return, subjects were contacted by telephone to verify survey receipt and to encourage survey completion. All patients who completed the 12-month survey by mail or phone received $10 compensation. Returned surveys were reviewed so that incomplete or inconsistent reporting could be followed up by contact with specific respondents. Completed 12-month surveys were electronically scanned. All baseline and follow-up survey questions were pilot-tested on a group of 20 postpartum adolescent mothers and revised. The questions were translated into Spanish using forward and backward translation; Spanish surveys underwent additional pilot testing.
The structured 12-month survey completed by the adolescent mothers elicited the following information: demographic and reproductive characteristics; tobacco, alcohol, and other drug use; intimate partner physical violence toward the adolescent mother; social support and isolation experienced; current partner's age and current partner's substance use.
All participants were asked if they had had sexual intercourse within the preceding 6 months. Those responding positively were then asked questions on contraception and condom use and whether they were pregnant or currently trying to conceive. Reliable contraception was affirmed if at last intercourse the adolescent mother reported the use of condoms, Norplant (Wyeth-Ayerst Laboratories, Philadelphia, Pa), the birth control pill, Depo-Provera (Pharmacia & Upjohn, Peapak, NJ), an intrauterine device, or a diaphragm. Frequency of condom use was measured by asking how often during the previous 6 months the adolescent mother's partner used a condom when they had sexual intercourse. Condom use response options were later dichotomized into 2 categories: "infrequent use" ("never," "sometimes," and "about half the time") and "frequent use" ("most times" and "every time"). Partner's refusal to use condoms or to allow the use of birth control were evaluated on a 4-point Likert scale and then dichotomized by collapsing "strongly agree" and "agree" into agree, and "strongly disagree" and "disagree" into disagree. An adolescent female was considered as having had a planned pregnancy or trying to conceive if she confirmed that her recent pregnancy (within the previous 11 months) was planned or that she was now trying to become pregnant. Rapid repeated pregnancy was defined as a pregnancy occurring in the interval between recruitment and the 12-month follow-up survey.
Intimate partner violence was affirmed if an adolescent mother's partner or husband hit her during an argument or while he was drunk or high at least once in the last 6 months. Questions on substance use by the adolescent mother and her perceptions of her partner's use measured tobacco, alcohol, marijuana, and other drug use. Daily tobacco use by the mother was confirmed if she reported smoking half a pack or more per day in the last 30 days. Use of other drugs included any use of amphetamines, barbiturates, tranquilizers, flunitrazepam, cocaine, inhalants, lysergide (LSD), or heroin, regardless of frequency. Partner's daily tobacco use and daily alcohol use were established if the adolescent mother reported that her partner smoked cigarettes daily or drank alcohol "every day" or "almost every day." Partners were considered weekly marijuana users if the adolescent mother reported they smoked marijuana or hashish at least 1 to 2 times per week. Partner's use of other drugs was established if the adolescent mother reported if any other drug (amphetamines, barbiturates, tranquilizers, flunitrazepam, cocaine, inhalants, LSD, or heroin) was used, regardless of frequency.
Social support was assessed in several ways. The Family Support and Overall Support scales used in this study grew out of research identifying common problems and/or stressors faced by adolescent mothers.16,17
Items for each scale were reviewed by experts for content validity, and the scales have high internal consistency. Adolescent mothers were asked to indicate on a 5-point scale from "never true"1 (1) to "always true" (5) the amount of emotional, financial, informational, transportation, and child care support received from family members. Sample items included "My family helps me with money when I need it," or "My family tells me I am a good mother." The resulting variable, Family Support, was calculated by summing all 7 items and dividing the total by 7. Family Support scores ranged from 1 to 5 with Cronbach α = .91. Overall support was evaluated by asking adolescent mothers to indicate, on the same 5-point scale, the amount of emotional, financial, informational, transportation, and child care support they might receive from anywhere. Sample items included "I get enough help with money when I need it" or "I get enough help with child care when I need it." All 5 items were summed and the total divided by 5 to produce Overall Support (Cronbach α = .85). For both sources of support, high scores indicate high support.
Social isolation was measured by asking the adolescent mother her frequency of contact with her own mother and with her friends on a 5-point scale from daily contact to contact less than once a month. Adolescent mothers were also asked whether they wished they could see their friends "more often," "less often," "about the same," or "not at all." For adolescent mothers living with an adult authority figure, this variable was defined as the adolescent's parents, grandparents, aunt/uncle, or guardian.
Age discrepancy between the adolescent mother and her partner was calculated by subtracting the adolescent mother's chronological age at 12 months' postpartum from the age of her current partner. To examine factors associated with older adult partners, patients were divided into 2 groups: those with older adult partners and those with similar-aged partners. These age cutoffs were used to be consistent with prior literature4,10,11
as well as the American Bar Association's age criteria for statutory rape.4 An older adult partner was defined as at least 5 years older than the adolescent mother and at least 20 years old (n = 184). A similar-aged partner was defined as ±2 years' age difference (n = 312). Couples with a 3- to 4-year age difference reflect societal norms10
and in many states do not meet the criteria for statutory rape based on age alone. Also, these couples contained adolescent mothers aged 14 to 18 years and therefore partners aged 17 to 22 years. It is unlikely that relationships between 14-year-old mothers and 18-year-old partners are comparable to those between 18-year-old mothers and 22-year-old partners. Analyses conducted using this "mixed" group would be difficult to interpret. Thus, the remaining girls (n = 339) with partners who were within 3 to 4 years of their own age were excluded from subsequent analyses.
Demographic characteristics, reproductive health behaviors, social support and isolation, substance use by the adolescent mother and her partner, and the occurrence of intimate partner violence were compared between the 2 groups using either χ2 analysis or t tests depending on the level of measurement and the extent to which parametric assumptions were met. Given the temporal ordering between the outcome variable (age discrepancy) and other variables of interest (drug use and current pregnancy), multiple logistic regression could not be used to examine potential modifying effects. Therefore, additional stratified analyses were conducted to evaluate the extent to which living with an adult authority figure (parent, grandparent, aunt/uncle, or other guardian) or being with the father of her infant born 12 months previously might alter these observed relationships. Criteria for statistical significance was set at P<.05. All analyses were conducted using statistical software (SPSS version 8.0; SPSS Inc, Chicago Ill).
Approximately 735 adolescent mothers (79%) completed surveys at 12 months' postpartum. Adolescent mothers who failed to return surveys did not significantly differ from the original cohort interviewed following delivery on demographic variables. Similar-aged partners averaged 18.8 ± 1.4 years of age (age range, 15-21 years); older adult partners averaged 25.5 ± 4.0 years of age (age range, 20-50 years) (P<.001).
Demographic characteristics and reproductive health between the 2 groups are given in Table 1. While adolescent mothers in both groups did not differ with respect to age, they were significantly more likely to report having achieved less than a ninth-grade education. Adolescent mothers with older adult partners were also less likely to be enrolled in school or to be employed. There were no differences between the 2 groups of adolescent mothers with respect to race/ethnicity and whether their current partner was the father of their infant born 12 months previously, the head of the household was employed 35 or more hours each week, there was enough money to live on, or there was enough transportation to go places.
Approximately 90% of all adolescent mothers reported having had sexual intercourse in the preceding 6 months (Table 1). Adolescent mothers with older adult partners were no more likely than mothers with similar-aged partners to have been pregnant since their infant's birth 12 months previously, to report seeing someone or becoming involved with someone other than their infant's father at 12 months' postpartum, or to report that any subsequent pregnancies were either unplanned or were with the man who fathered their earlier child. However, mothers with older adult partners were more likely to report that a subsequent pregnancy was planned or that they were currently trying to conceive (Table 1). This is consistent with their report of infrequent condom use or that their partners refused to use condoms. However, adolescent mothers with older adult partners were no less likely to report using reliable contraception or to affirm that their partners would not allow them to use birth control.
Contrary to our hypothesis, there were no differences between the 2 groups regarding intimate partner violence experienced during the preceding 6 months or the adolescent mother's or partner's use of tobacco, alcohol, marijuana, or other drugs over the preceding 30 days (Table 2). However, daily alcohol use was more common among the older adult partners than among the similar-aged partners.
Several differences were found between the groups of adolescent mothers regarding social support and isolation experienced (Table 3). Adolescent mothers with older adult partners were more likely to be living on their own, away from an adult authority figure, despite the fact that they were no more likely to be married or living with this partner. While these adolescent mothers had less contact with their own mothers, they reported receiving less family support and less overall support than did adolescent mothers with similar-aged partners. No differences were noted between groups compared with respect to frequency of contact with friends.
To evaluate the possible mediating effects of living with an adult authority figure on the behaviors and experiences of adolescent mothers in both groups, data were stratified according to whether or not they were living with an adult authority figure. Only comparisons significant at P<.05 are reported in Table 4. Of interest, 101 (46.1%) of 219 adolescent mothers not living with an adult authority figure reported being partnered with an older adult compared with 82 (29.7%) of 276 living with an adult authority figure (P<.001).
With respect to demographic characteristics, adolescent mothers with older adult partners not living with an adult authority figure were younger than adolescent mothers with similar-aged partners who did not live with an adult authority figure and were more likely to be Mexican American (Table 4). Adolescent mothers with older adult partners who did not live with an adult authority figure achieved less education and were less likely to be enrolled in school, to be employed, or to report that they had enough money to sustain themselves. Groups compared did not differ with respect to having completed high school, General Educational Development test, or vocational degrees.
Regarding reproductive behaviors, adolescent mothers with older adult partners were more likely to report a recent planned pregnancy or an attempt to conceive, regardless of their living situation (Table 4). However, adolescent mothers with older adult partners who were not living with an adult authority figure were more likely to report that their current partners refused to use condoms or would not allow the use of other birth control methods compared with those with similar-aged partners. In the groups compared, adolescent mothers living with or not living with an adult authority figure did not differ with respect to having experienced a planned repeated pregnancy, use of reliable methods of birth control, and frequency of condom use.
Adolescent mothers with older adult partners reported higher rates of partner daily alcohol use only when adolescent mothers were living with an adult authority figure (Table 4). Additionally, adolescent mothers with similar-aged partners reported higher rates of other drug use only when the adolescent mothers were not living with an adult authority figure. The percentages of adolescent mothers reporting intimate partner violence or their partners' use of tobacco or other drugs did not differ between groups regardless of living status.
Infrequent contact with her own mother was reported more often by adolescent mothers with older adult partners only for those not living with an adult authority figure; these adolescent mothers also reported lower levels of overall support. None of the groups compared differed with respect to contact with friends, living with their partners, or mean support received from family members.
To evaluate the possible mediating effects of the blood relationship of the current partner to the infant born 12 months previously on the behaviors and experiences of adolescent mothers in both groups, the data were stratified according to whether or not the current partners of the adolescent mothers fathered their earlier infant(s). Only those statistical comparisons significant at P<.05 are given in Table 5. Adolescent mothers with older adult partners were significantly younger than those with similar-aged partners only if their current partner was the father of their infant born 12 months previously. Adolescent mothers with older adult partners achieved a lower level of education and were less likely to be enrolled in school or employed only when their current partner also fathered their previous infant(s). The groups compared did not differ with respect to adolescent mothers' race/ethnicity or any of the financial indicators.
Adolescent mothers with older adult partners were more likely to report a recent planned pregnancy or attempting to conceive and that their partner refused to use condoms only when those partners were not the fathers of their previous infant(s). No differences were noted between groups in having experienced a subsequent pregnancy or the use of reliable birth control.
The percentages of adolescent mothers reporting intimate partner violence or their use of alcohol, tobacco, marijuana, or other drugs did not differ between groups regardless of whether their partners fathered their earlier infant(s). However, daily alcohol use and weekly marijuana use by the partner was reported more frequently by adolescent mothers with older adult partners only when their current partner did not father their previous child.
Finally, social support and isolation between age-discrepant pairings were compared within groups of partners who were and were not the fathers of infants born 12 months previously. Only adolescent mothers with older adult partners who were also the fathers of their previous infant(s) were less likely to report living with a parent or guardian (Table 5). Similarly, these mothers were more likely to report infrequent contact with their own mothers, lower family support, and lower overall support. The groups compared did not differ with respect to whether they were living with their partners or frequency of contact with friends.
Consistent with reports of pregnant adolescents,2,6
we found that 1 in 5 adolescent mothers report having a partner who is both an adult and at least 5 years older at 12 months' postpartum. The absence of racial/ethnic differences in the prevalence of partnering with an older adult male observed in this study has been reported by others.1,3,6,18
Thus, the preference by adolescent mothers for older adult partners does not seem to be unique to one particular race/ethnicity.
Adolescent mothers with older adult partners were more likely to have experienced a subsequent planned pregnancy or were currently trying to conceive, findings consistently observed across both sets of stratified analyses. Thus, a planned repeated pregnancy is more likely for adolescent mothers with older adult partners regardless of whether they currently live with a parent or guardian or whether their current partner also fathered their previous child. Through informal conversations, some study participants revealed that they were trying to conceive to maintain the romantic interest of the father of their previous child or to provide a child to a new partner. While planned pregnancies are preferable to unplanned pregnancies, additional childbearing may place a young mother at increased risk for financial hardship should her relationship dissolve.19,20
We did not find that adolescent females with older adult partners were more likely to experience intimate partner violence. These observations held across both sets of stratified analyses and are consistent with findings by Rickert et al6 who observed that there were no differences in intimate partner violence between pregnant adolescents with similar-aged vs older adult partners. Given the unequal power balance often found in relationships in which one partner is significantly older, more educated, or has greater life experiences than the other, we also examined more subtle forms of possible coercion and control, including partner's refusal to use condoms or contraception. Across both sets of stratified analyses, adolescent mothers with older adult partners were more likely to report that their partners refused to use condoms. However, only adolescent mothers with older adult partners who were not living with a parent or guardian were more likely to report that their partner did not allow the use of contraception. Because young mothers often lack the emotional and cognitive development or experiential background necessary to recognize or control potential coercive strategies, they may be unable to negotiate the use of condoms or contraception. These findings may help explain the increased risk of a planned repeated pregnancy among adolescent mothers with older adult partners observed here.
Young mothers with older adult partners in this sample were more likely to have established independent households and to receive limited support from their families. It is possible that family support may have been withdrawn in direct response to their selection of an older adult partner. Alternatively, adolescent mothers in this study who received little family support may have sought a relationship with an older adult partner in an effort to find financial and emotional security. Prior research has demonstrated that adolescent mothers who receive low levels of family support experience more stress and are less likely to complete school than adolescent mothers with high levels of family support.12 However, only adolescent mothers with older adult partners who either were not also living with an adult authority figure or whose current partner also fathered their previous child were less likely to be in school or employed and had completed less formal education. Hence, it appears that continued residence with family members may provide an important role in mediating the effects of being with an older adult partner on educational and financial outcomes. Given the high rates of relationship dissolution observed among adolescent mothers with partners of any age,19 young mothers no longer living with their families and their children may also be at extreme risk of long-term financial disadvantage. However, further longitudinal research is required to better evaluate the financial sequelae of having an older adult partner.
Contrary to our hypothesis, we did not find that the use of alcohol and other drugs differed between groups despite that older adult partners, especially those who were not the father of the infant born 12 months previously, were significantly more likely to drink alcohol or use marijuana on a weekly or more frequent basis. These findings held across both sets of stratified analyses and stand in contrast to the strong association previously reported between partner alcohol and other drug use and substances used by pregnant adolescents.6,21,22
Thus, some young women may use early motherhood as an opportunity to create positive change in their lives and so experience an accelerated transition from a risk-taking adolescent into more responsible adult roles.
Several limitations to this study bear mentioning. First, most participants were from lower socioeconomic levels and lived in southeast Texas. Thus, findings from this study may not generalize to adolescent mothers from higher socioeconomic levels, different race/ethnic groups, or different geographic regions. Also, all information reported was obtained from the adolescent mother. Direct questioning of male partners may reveal different patterns of results. Although the social support variables used in this study have been evaluated for content validity and demonstrate strong internal consistency, these scales require further validation. Finally, this was a largely descriptive study, and a prospective study is needed to further examine relationship dynamics and life circumstances of adolescent mothers with older adult partners.
In summary, adolescent mothers with older adult partners were no more likely to use alcohol or other drugs or to experience intimate partner violence. The negative educational and financial impact of coupling with an older vs similar-aged partner was greater for adolescent mothers who no longer resided with an adult authority figure. These adolescent mothers were also at greater risk of a planned rapid repeated pregnancy. Given their limited educational attainment and family support, a subsequent pregnancy could easily place these mothers and their offspring at considerable financial and educational disadvantages. Additional studies are needed to determine the longer-range effects of these pairings on adolescent mothers' subsequent educational and financial success. Further research also needs to examine more closely the relationship factors that could explain the cause of a planned repeated pregnancy.
Accepted for publication January 29, 2001.
This research was supported by grants from the National Institute of Health and Drug Abuse, Bethesda, Md (DA09636, DA08404), and the Hogg Foundation for Mental Health, Austin, Tex (Dr Wiemann). Ms Agurcia is a National Institutes of Health, Bethesda, Minority Supplement appointee (DA09636-S1).
Presented in part at the Annual Meeting of the Society for Adolescent Medicine, Los Angeles, Calif, March 18, 1999.
We thank the many adolescent participants who shared their experiences with us and the interviewers who spent hundreds of hours collecting this information.
Corresponding author and reprints: Constance M. Wiemann, PhD, Adolescent Medicine and Sports Medicine Section, Department of Pediatrics, Baylor College of Medicine, One Baylor Plaza, Houston, TX 77030 (e-mail: firstname.lastname@example.org).
Agurcia CA, Rickert VI, Berenson AB, Volk RJ, Wiemann CM. The Behavioral Risks and Life Circumstances of Adolescent Mothers Involved With Older Adult Partners. Arch Pediatr Adolesc Med. 2001;155(7):822-830. doi:10.1001/archpedi.155.7.822