Adolescents at risk for mistreating their children

Pergamon Child Abuse & Neglect 6 (2001) 737–751

Adolescents at risk for mistreating their children Part I: prenatal identification Catherine Stevens–Simon*, Donna Nelligan, Lisa Kelly Department of Pediatrics, Division of Adolescent Medicine, University of Colorado Health Sciences Center, The Children’s Hospital, 1056 E. 19th Street, Denver, CO 80218, USA Received 2 August 2000; received in revised form 20 October 2000; accepted 31 October 2000

Abstract Objective: To determine if the Family Stress Checklist helps prenatal care providers identify adolescents who are at risk for mistreating their children. Methods: We studied 262 participants in a comprehensive, adolescent-oriented maternity program. During the prenatal period, the Family Stress Checklist was used to quantify abuse potential, with scores ⬎25 defining high risk. Information about the social context of the pregnancy and the pattern of health care utilization was obtained with a self-administered questionnaire, and by reviewing the medical records. Major disruption of primary care giving by the adolescent mother was classified hierarchically as abuse, neglect, and abandonment. Results: Family Stress Checklist scores ranged from 0 to 65 (mean ⫹ SD ⫽ 20.1 ⫹ 1.4); 113 (43%) of the 262 teenagers were classified as high risk. High and low risk adolescent mothers made an equivalent number of health maintenance and Emergency Department visits, but the high risk group initiated significantly more acute care visits (6.0 ⫹ 4.1 compared to 3.9 ⫹ 3.3; p ⬍ .0001). After controlling for pre-existing sociodemographic differences, high risk 1-year-olds were 8.41 (95% CI:1.77– 40.01) times and high risk 2-year-olds 5.19 (95% CI:1.99 –13.60) times more likely to have been mistreated than their low risk counterparts. Conclusions: Prenatal care providers can use the Family Stress Checklist to systematically identify a subgroup of adolescent mother whose excessive use of the acute medical care services and propensity for mistreating their children suggests the need for additional support services. © 2001 Elsevier Science Ltd. All rights reserved. Keywords: Child abuse; Adolescent pregnancy; Adolescent parenting

* Corresponding author. 0145-2134/01/$ – see front matter © 2001 Elsevier Science Ltd. All rights reserved. PII: S 0 1 4 5 - 2 1 3 4 ( 0 1 ) 0 0 2 3 6 - 8

738

C. Stevens–Simon et al. / Child Abuse & Neglect 6 (2001) 737–751

Introduction Age-related differences in the prevalence of personal and environmental risk factors for dysfunctional parenting, and cognitive and psychosocial maturity substantially increase the risk of child abuse and neglect in adolescent-headed households (Flanagan, Coll, Andreozzi, & Riggs, 1995; Flanagan, McGrath, Meyer, & Garcia Coll, 1995; Stevens–Simon & Reichert, 1994; Stier, Leventhal, Berg, Johnson, & Mezger, 1993; Zuravin, 1988). Although the actual prevalence of child abuse and neglect in the United States is unknown, it is estimated that 20% to 30% of American adults were abused as children, and that the children of adolescent mothers are twice as likely to be abused as the children of adult mothers (Stevens–Simon & Reichert, 1994; Stier et al., 1993). However, most adolescent parents do not mistreat their children (Flanagan, 1995; Stevens–Simon & Reichert, 1994; Stier et al., 1993; Zuravin, 1988). Because interventions designed to prevent the maltreatment of children are costly and most effective with parents who are at risk for abusive and neglectful care-giving, the feasibility of predicting negative parenting behavior on the bias of information collected during the perinatal period has been extensively investigated (Brayden et al., 1993; Flanagan et al., 1995; Leventhal, Garber, & Brady, 1989; Murphy, Orkow, & Nicola, 1985; Olds & Kitzman, 1990; Stevens–Simon & Reichert, 1994; Zuravin, 1988). The consensus is that it is possible to identify parents who are at risk for mistreating their children, and that the benefits of preventing child abuse and neglect outweigh the adverse consequences of overdiagnosis (Brayden, Altemeier, Dietrich, Tucker, Christensen, McLaughlin, & Sherrod, 1993; Leventhal et al., 1989; Murphy et al., 1985). A previous study conducted at this institution demonstrated that The Family Stress Checklist accurately identifies neonates who are at risk for abuse and neglect (Murphy et al., 1985). The purpose of this study was to determine if prenatal health care and social service providers could use the Family Stress Checklist to systematically identify a subgroup of the participants in an adolescent-oriented antenatal program who would be more likely to benefit from an intensive, home-based, parenting intervention.

Methods Subjects The study sample consisted of a racially and ethnically diverse (45% White, 32% Black, 22% Hispanic, and 1% other) group of 262 poor (92% Medicaid recipients), predominantly unmarried (94%), primiparous (98%), 13 through 19-year-old mothers (mean ⫹ SD ⫽ 16.6 ⫹ 1.4 years) who: (1) obtained prenatal care in the Colorado Adolescent Maternity Program (CAMP); (2) gave birth to normal, singleton infants at the University of Colorado Hospital between January 1, 1994 and December 31, 1996; and (3) identified CAMP as their primary health care provider 2 months after delivery. The last restriction was added because only approximately 80% of CAMP prenatal patients join the postpartum program (Stevens– Simon, Kelly, & Kulick, 2000). Although lack of transportation and the desire to utilize their own former pediatric health care provider are the most frequent reasons teenagers give for

C. Stevens–Simon et al. / Child Abuse & Neglect 6 (2001) 737–751

739

leaving CAMP during the puerperium, those who leave the program are not a random sample of the CAMP prenatal population (Stevens–Simon et al., 2000). Rather, as a group, they tend to be younger, to be White, to have less adequate social support, and to score higher on The Family Stress Checklist (Stevens–Simon et al., 2000). Because White CAMP patients are the least apt to visit the Emergency Department, this study could overestimate Emergency Department use. In addition, the selective loss of teenagers with high Family Stress Checklist scores could result in an underestimation of the incidence of child maltreatment in the CAMP prenatal population. CAMP patients who were over 19 years of age at delivery were excluded from this analysis because most young women graduate from high school by the time they are 19 years old, making this age a natural cutoff between adolescence and young adulthood. We also excluded 14 mother-infant dyads whose health status could not be tracked for at least 6 months after delivery. There was no significant difference between the Family Stress Checklist scores of the 14 excluded teenagers and 262 teenagers who were included in this analysis (mean ⫹ SD ⫽ 15.4 ⫹ 10.5 and 20.1 ⫹ 12.4, respectively). The study was approved by the Institutional Review Board at the University of Colorado Health Sciences Center. The Colorado Adolescent Maternity Program CAMP is a comprehensive, multidisciplinary prenatal, delivery, and postnatal care program located in a large, urban, teaching hospital (Stevens–Simon et al., 2000). The program integrates the professional services of health care providers trained in obstetrics, pediatrics, and adolescent medicine, a social worker, and a dietician. The goal is to reduce the incidence of adverse pregnancy outcomes and repeat teen pregnancies, and to increase the number of young people who graduate from high school, and become productive members of their communities and nurturing, nonabusive parents. To promote staff-patient interaction and facilitate the implementation of individual care plans following delivery, parents and children are seen together and appointments are scheduled monthly for the first 6 months, every other month for the next 6 months, and then at 3 month intervals until the child is 2 years old. Thus, nine health maintenance visits are scheduled during the first postpartum year, and four during the second. The components of the postpartum program were selected with the expectation that the intervention would prevent adverse maternal and child outcomes directly by simplifying access to preventative health care and social services, and indirectly by discouraging school drop-out, encouraging the pursuit of careers that foster competency and are incompatible with closely spaced adolescent childbearing, enhancing family support, and promoting linkages to community service organizations. At each health maintenance visit, providers emphasized the advantages of delaying further childbearing beyond adolescence, and tried to identify and counter misunderstandings about child development, disciplinary styles, and childrearing attitudes that set the stage for child abuse and neglect. Families met with the social worker and dietician when necessary. To teach the young parents how to provide a safe, stimulating home environment, at each health maintenance visit they were given safety tips for their home, and a set of age-appropriate, development-promoting games to play with their children. Additional measures implemented to eliminate common barriers to care included: (1) a waiting time for health and contraceptive maintenance appointments of less than one

740

C. Stevens–Simon et al. / Child Abuse & Neglect 6 (2001) 737–751

week; (2) a telephone or mail recall system conditioned on missed health maintenance visits, with rescheduling of missed appointments within one week of contact; (3) special efforts to schedule appointments at times which did not conflict with the parents’ school and/or work schedules; (4) unscheduled walk-in visits available daily; (5) clinic fees, immunizations, and contraceptive supplies offered on a sliding scale, with free services and supplies for uninsured and underinsured patients; (6) bus tokens and help accessing other forms of free transportation; and (8) a policy of routinely administering needed vaccines to children with minor acute illnesses, and needed contraceptives to teenagers at scheduled and unscheduled well and sick visits (Stevens–Simon et al., 2000). When the clinic was closed, CAMP patients were to use an after-hours, nurse call-in line before going to the Emergency Department (Kempe, Dempsey, Whitefield, Botherner, MacKenzie, & Poole, 2000). Data collection and definition of variables As part of the enrollment procedure, all CAMP prenatal patients met with the social worker and completed a precoded, self-administered, multiple-choice questionnaire. The Family Stress Checklist was administered during the initial social worker assessment. To this end, specific questions designed to elicit information bearing on the 10 checklist items were interspersed in the intake interview. Each item of the Family Stress Checklist was scored as: 0-no risk; 5-risk; and 10-high risk, and the 10 items were summed to yield a total scale score ranging from 0 to 100 (Murphy et al., 1985). The intake questionnaire was written at a fourth grade reading level, and was designed to collect information about the social context of the pregnancy, with emphasis on commonly cited demographic, psychosocial, and behavioral risk factors for low birth weight and preterm delivery, repeat pregnancy, and child maltreatment during adolescence (Stevens–Simon et al., 2000; Stevens–Simon & White, 1991). Additional medical and outcome data were obtained by reviewing the participants’ medical records. Independent variable Abuse-risk group: the Family Stress Checklist score was treated as a continuous and a categorical variable. For categorical analyses, the score was dichotomized at 25 because a prior study (Murphy et al., 1985) found that by 2 years of age 25% of the children born to adult mothers who scored 25 or more on the Checklist had been reported to the Department of Social Services for maltreatment, compared to only 2% of the children whose mothers scored under 25 on the Checklist. Dependent variables The maltreatment of children is a multidimensional problem which has been broadly defined as any parenting behavior that has deleterious effects on the physical and psychological growth and development of children (Wissow, 1995). Although these negative care-giving practices are not mutually exclusive, and many infants and children suffer more than one type of maltreatment, we classified the dysfunctional parenting behaviors that

C. Stevens–Simon et al. / Child Abuse & Neglect 6 (2001) 737–751

741

resulted in a major disruption of primary care-giving by the adolescent mother, hierarchically, according to the following criteria: (1) abuse including physical (inflicting bodily harm through excessive force or forcing a child to engage in physically harmful activities), sexual (contact or interaction between a child and an adult, when the child is being used for the sexual stimulation of that adult or another person), and emotional (coercive, demeaning, or overly distant behavior by a caretaker that interferes with a child’s normal social or psychological development); (2) neglect (failure to provide basic shelter, supervision, medical care, or support for a child); and (3) abandonment (the mother left the home, putting a friend or relative in charge of the child’s care). In addition to these traditional forms of child maltreatment, we recorded the timing of the next pregnancy. Although repeat conception is not usually an endpoint in studies of child maltreatment, data compiled over the last two decades suggest that excessive environmental and personal stress that is unmitigated by social support promotes dysfunctional, negative parenting behavior (Flanagan, 1995; Olds & Kitzman, 1990; Stevens–Simon & Lowy, 1995; Stevens–Simon & Reichert, 1994; Zuravin, 1988; Wissow, 1995). The stress associated with a rapid repeat conception has been identified as a particularly important mediator of the relationship between adolescent parenthood and child abuse and neglect (Zuravin, 1988). Finally, because studies suggest that the frequency of parent-initiated pediatric acute care visits identifies mothers and infants who are experiencing a strain in their relationship (Hardy & Street, 1989; Harris, Weston, & Lieberman, 1989), and because such visits represent an expensive, potentially preventable, health care expenditure, information about the pattern of health care utilization (e.g., the number of CAMP clinic visits for sick and well infant and teen care, Emergency Department visits, and hospitalizations) was abstracted from the medical records. We also recorded the timing of immunizations, since failure to comply with the recommended schedule for childhood immunizations (US Department of Health and Human Services, Public Health Service, 1992) is a type of medical neglect. Intervening variables Variables from several domains affect the risk of child maltreatment. The specific variables examined as potential confounders of the relationship between the Family Stress Checklist score and subsequent child maltreatment were identified a priori from a review of the adolescent parenting and child maltreatment literatures (Brayden et al., 1993; Flanagan, 1995; Flanagan, 1995; Leventhal et al., 1989; Murphy et al., 1985; Olds & Kitzman, 1990; Stevens–Simon et al., 2000; Stevens–Simon & Lowy, 1995; Stevens–Simon & Reichert, 1994; Stevens–Simon & White 1991; Stier et al., 1993; Wissow, 1995; Zuravin, 1988). For this study, only data obtained during the prenatal period were analyzed. Specifically, the variables of interest included: (1) sociodemographic factors such as age, race, and type of health insurance; (2) psychosocial factors such as living arrangements, school enrollment, involvement in socially problematic behaviors (fighting, delinquency, and illicit substance abuse, assessed by history and periodic urine toxicology screening), past or present physical or sexual abuse, suicide attempts, depression/stress [defined by the Center for Epidemiologic Studies Depression scale and a checklist of 20 negative life events, adapted from Newton’s scale of psychosocial stressors in pregnancy (Newton, Webster, Binu, Maskrey, & Phillips,

742

C. Stevens–Simon et al. / Child Abuse & Neglect 6 (2001) 737–751

1979; Radloff, 1977; Stevens–Simon & McAnarney, 1995)]; lack of support from prime people, [e.g., their own mother or the baby’s father (Barnet, Joffe, Duggan, Wilson, & Repke, 1996)]; and inadequate family support [defined by the Family Apgar Scale (Smilkstein, Ashworth, & Montano, 1982)]; and (3) pregnancy-related factors such as the wantedness of the pregnancy, the birth weight and gestational age of the infant, and neonatal intensive care unit admissions. Data analysis Summary statistics were used to describe the study population. Comparisons between the two abuse-risk groups were conducted with Students t tests when the outcome variable was continuous, and ␹2 analyses when the outcome variable was categorical. The sensitivity, specificity, and positive and negative predictive values of the Family Stress Checklist for the various types of maltreatment were calculated. Finally, multivariate analyses using logistic regression were conducted to determine whether findings at the bivariate level would be supported after adjusting for pre-existing group differences in sociodemographic characteristics, with a hypothesized relationship to dysfunctional parenting behavior. To simplify the model and its application in clinical practice, the intervening variables were dichotomized (“present” or “absent”). Because we had no basis for choosing one variable over another, intervening variables for which there were significant abuse-risk group differences at the bivariate level were allowed to enter the forward, step-wise, logistic regression models one at a time, on the basis of the statistical significance of their association with the outcome variable of interest. Adjusted odds ratios and their 95% confidence intervals were calculated from the logistic coefficients and standard errors for each variable. The statistical test of the model was the ␹2 likelihood ratio. Statistical analyses were performed with SPSS/PC⫹ (Nie, Hull, Jenkins, 1989).

Results Scores on the Family Stress Checklist ranged from 0 to 65 (mean ⫹ SD ⫽ 20.1 ⫹ 1.4). The high risk group included 113 (43%) of the 262 teenagers; their Checklist scores ranged from 25 to 65 (mean ⫹ SD ⫽ 31.7 ⫹ 8.0). The remaining 149 teenagers made up the low risk group; their Checklist scores ranged from 0 to 20 (mean ⫹ SD ⫽ 11.2 ⫹ 6.1). Selected demographic and psychosocial characteristics of the 113 high risk and 149 low risk teenage mothers are compared in Table 1. The differences were not unanticipated, since parents receive points on the Family Stress Checklist for living in chaotic, unsupportive environments, being stressed and depressed, and engaging in antisocial behavior. Indeed, because the Family Stress Checklist actually includes items that require the interviewer to assess the constructs tapped by the stress, depression, and social support scales, we did not use these scores as independent predictor variables in subsequent multivariate analyses. By contrast, because the decision to live with a relative and the decision to remain in school could be a reflection of either the overall level of environmental chaos (a construct tapped by the Checklist), or intrinsic, personal characteristics of the adolescent mother (which are not

C. Stevens–Simon et al. / Child Abuse & Neglect 6 (2001) 737–751

743

Table 1 Characteristics of high and low risk adolescent mothers Characteristics

Sociodemographic: N(%) Conceived prior to age 16 years Minority race/ethnicity Food shortage at home⫹ Psychosocial: N(%) Not living with a parent Not living with a related adult School drop-out School problems* Problem behaviors** Depressed/stressed⫹ No prime support Poor family support⫹ Positive urine toxicology***⫹⫹

High

Abuse risk group Low

Significance (P)

(N ⫽ 113) 36 (32) 50 (44) 31 (28)

(N ⫽ 149) 42 (28) 92 (62) 25 (17)

ns .005 .04

70 (62) 54 (48) 62 (55) 91 (81) 82 (73) 69 (62) 29 (26) 42 (38) 18 (17)

65 (44) 52 (35) 52 (35) 90 (60) 49 (33) 55 (38) 13 (9) 26 (18) 12 (8)

.003 .03 .001 ⬍.0001 ⬍.0001 ⬍.0001 ⬍.0001 ⬍.0001 .04

* Truancy, fighting, suspension, expulsion ** Substance abuse, legal problems, rape, suicide, runaway *** All positive screens contained metabolites of marijuana Missing data for high and low risk groups: ⫹ N ⫽ 111 and 147;

⫹⫹

N ⫽ 104 and 145, respectively.

assessed by the Checklist, but could affect her ability to parent), these two variables were included in subsequent multivariate analyses. The two groups of teenagers had similar obstetrical histories; there were no significant group differences in gravity or parity; however, those classified as high risk were significantly more apt to have experienced a miscarriage (14.2% compared to 6%; p ⫽ .04). The proportion of wanted pregnancies, the timing of the first prenatal visit, compliance with prenatal appointments, gestational age at delivery, and the proportion of infants admitted to the neonatal intensive care unit was also similar in the two groups. Approximately one third of the 262 teenagers admitted that they had not been using birth control before conception, in part because they wanted to be or at least did not mind becoming pregnant; 12% received an inadequate amount of prenatal care, by the standards of the Institute of Medicine (Kessner, Singer, & Kalk, 1973); 7% gave birth prematurely (before the 37th week of gestation); and 7% of the newborns were admitted to the neonatal intensive care unit. Despite these similarities, the infants of high risk teenagers weighed less at birth than the infants of low risk teenagers (mean ⫹ SD ⫽ 3062 ⫹ 503 g compared to 3196 ⫹ 475 g; p ⫽ .03), and were twice as likely to be low birth weight (weight less than 2500 g at birth; 11.5% and 5.4%, respectively). Of the 262 mother-infant dyads initially enrolled in this study, 238 (90.8%; 86.7% of the high risk teenagers and 94.0% of the low risk teenagers) identified CAMP as their primary health care provider for at least 6 months after delivery (p ⫽ ns), and 186 (71.0%; 68.1% of the high risk teenagers and 73.2% of the low risk teenagers) did so for the first postpartum year (p ⫽ ns). The data presented in Table 2 show that marked group difference in the pattern of health care utilization emerged during this time. High and low risk adolescent mothers

744

C. Stevens–Simon et al. / Child Abuse & Neglect 6 (2001) 737–751

Table 2 Health care visits during the first postpartum year Visit type

Abuse risk group High

Low ⫹

CAMP Clinic [Number (mean ⫹ sd)] Well visits⫹⫹ Sick visit* Emergency Department visits⫹⫹⫹ Number (mean ⫹ sd) Made at least 1 visit: N(%) Hospitalized N(%)⫹⫹⫹⫹

(N ⫽ 77)

(N ⫽ 109)

6.1 ⫹ 2.1 6.0 ⫹ 4.1

5.9 ⫹ 2.3 3.9 ⫹ 3.3

1.3 ⫹ 1.4 47 (61) 4 (5)

1.6 ⫹ 2.1 69 (63) 4 (4)

⫹

186 (71%) of the 262 mother-infant dyad identified CAMP as their primary health care provider for at least 1 year; 77 (68%) of the 113 high-risk teens and 109 (73%) of the 149 low-risk. ⫹⫹ CAMP patients have 9 scheduled well visits during the first year; kept proportion ⫽ 66%. ⫹⫹⫹ Overall may be falsely elevated due to the selective loss of White CAMP patients (see text). ⫹⫹⫹⫹ Hospitalized at least once, all for acute illnesses. * p ⬍ .0001.

made an equivalent number of health maintenance and emergency department visits, but members of the high risk group initiated significantly more acute care CAMP clinic visits. There was no statistically significant group difference in the incidence of underimmunization during the first two postpartum years. Overall, 28% of 9-month-olds, 16% of 15-month-olds, and 6% of 2-year-olds were underimmunized, by the standards of the American Academy of Pediatrics (US Department of Health and Human Services, Public Health Service, 1992). High risk infants were more likely to be admitted to the hospital than low risk infants during the first 6 months of life (5.1% compared to .7%; p ⫽ .06), but not thereafter. The repeat pregnancy rate was also similar in the two groups, with 14% of the teenagers becoming pregnant again during the first postpartum year and 35% during the second. During the first 6 postpartum months, 7 (2.7%) of the 262 infants were removed from their mother’s custody, 2 (.8%) were abused, 3 (1.1%) were neglected, and 2 (.8%) were abandoned. Although there were no statistically significant group differences in the incidence of maltreatment, both cases of abuse occurred in the high risk group. During the second half of the first postpartum year, 4 (3.5%) of the 113 infants in the high risk group and 9 (7%) of the 149 infants in the low risk group were lost to follow-up (p ⫽ ns). The data presented in Table 3 show that infants who were identified during the perinatal period as being at increased risk for subsequent abuse and neglect were significantly more likely to have been mistreated and removed from their mother’s custody than their low risk counterparts. Moreover, none of the infants in the low risk group were abused during the first postpartum year. A multiple logistic regression analysis examining the relationship between Family Stress Checklist score and child maltreatment during the first year of life, while controlling for antecedent group differences in race, living arrangements, and school status identified membership in the high risk group as the only statistically significant, independent predictor of maltreatment (RR ⫽ 8.41; 95% CI:1.77– 40.01; model ␹2 ⫽ 11.07; p ⫽ .0009). The

C. Stevens–Simon et al. / Child Abuse & Neglect 6 (2001) 737–751

745

Table 3 Relative risk of maltreatment during the first year of life for high and low risk group infants Type of maltreatment⫹

Incidence

RR

95% CI

P

– 5.21 6.58 8.41

–

– ns ns .006

Risk Group High N(%) Abuse Neglect Abandonment Any maltreatment

(N 3 4 5 12

Low

⫽ 109) (2.8) (3.7) (4.6) (11.1)

⫹⫹

(N ⫽ 140) 0 1 (0.7) 1 (0.7) 2 (1.4)

.55–48.91 .73–59.15 1.77–40.01

⫹

Dysfunctional parenting behaviors that resulted in a major disruption of primary caregiving by the adolescent mother, were classified hierarchically. Forward step-wise logistic regression models, all of which controlled for pre-existing group differences in race, living arrangements, and school status. ⫹⫹ During the second half of the first postpartum year, 4 (3.5%) of the 113 infants in the high-risk group and 9 (7%) of the 149 infants in the low-risk group were lost to follow-up (p ⫽ ns).

sensitivity of a prediction of maltreatment was 85.7%, but the specificity was only 58.4%, with corresponding positive and negative predictive values of 11% and 98.6%, respectively. Data through the end of the second postpartum year was available on 214 (81.7%) of the 262 mother-infant dyads: 99 (87.6%) of the 113 high risk and 115 (77.2%) of the 149 low risk pairs (p ⫽ .02). Table 4 shows that group differences in maltreatment persisted during the second postpartum year. By the end of the study, 22 (22.2%) of the 99 infants born to these high risk adolescent mothers had been maltreated and were as a result either voluntarily or forcibly estranged from their mother, compared to only 6 (5.3%) of the 115 infants of low risk adolescent mothers (p ⬍ .0001). We used the same multivariate model to examine the Table 4 Relative risk of maltreatment during the first two years of life for high- and low-risk group infants Type of maltreatment⫹

Incidence

RR

95% CI

P

2.38 4.97 6.35 5.19

.41–13.87 1.00–24.53 1.31–30.88 1.99–13.60

ns .05 .02 .0007

Risk Group High N (%) Abuse Neglect Abandonment Any maltreatment

(N ⫽ 99) 4 (5) 8 (8) 10 (10) 22 (22)

Low ⫹⫹

(N ⫽ 115) 2 (1.7) 2 (1.7) 2 (1.7) 6 (5.2)

⫹ Dysfunctional parenting behaviors that resulted in a major disruption of primary caregiving by the adolescent mother were classified, hierarchically; however these negative caregiving practices were not mutually exclusive, and many infants suffer more than one type of maltreatment. Ultimately, 9 (9%) of the 99 high-risk infants and 2 (1.7%) of the 115 low-risk infants were neglected (RR: 5.65 95% CI: 1.16 –27.39; p ⫽ 03,) and 13 (13%) of the high-risk and 2 (1.7%) of the low-risk infants were abandoned (RR: 8.54; 95% CI: 1.82–39.65; p ⫽ 006). Forward step-wise logistic regression models, all of which controlled for pre-existing group differences in race, living arrangements, and school status. ⫹⫹ During the second postpartum year, 14 (12.4%) of the 113 infants in the high-risk group and 34 (22.8%) of the 149 infants in the low-risk group were lost to follow-up (p ⫽ .02).

746

C. Stevens–Simon et al. / Child Abuse & Neglect 6 (2001) 737–751

relationship between Checklist score and dysfunctional parenting behavior during the second postpartum year, and again found that membership in the high risk group was the only statistically independent predictor of maltreatment (RR ⫽ 5.19; 95% CI:1.99 –13.60; model ␹2 ⫽ 14.06; p ⫽ .004). The sensitivity and specificity of a prediction of maltreatment during the first 2 postpartum years were 78.6% and 41.4%, respectively, with corresponding positive and negative predictive values of 22.2% and 94.8%.

Discussion Women who begin childbearing during their teens are over-represented in child maltreatment cases, and tend to take a more punitive approach to childrearing, to cope less effectively with the stresses of child care, and to be less able to protect their children from physical trauma than older, more cognitively and emotionally mature women (Flanagan et al., 1995; Flanagan et al., 1995; Stevens–Simon & Reichert, 1994; Stier et al., 1993; Zuravin, 1988). Despite the well documented difficulties associated with childrearing at this age, the results of this and most other studies demonstrate that the majority of adolescent parents do not need costly, home-based services to avoid mistreating their children (Flanagan et al., 1995; Murphy et al., 1985; Olds, Eckenrode, Henderson, Kitzman, Powers, Cole, Sidora, Morris, Pettitt, & Luckey, 1997; Olds, Henderson, Chamberlin & Tatelbaum, 1986; Stevens–Simon & Reichert, 1994; Stier et al., 1993; Zuravin, 1988). Rather, we found that prenatal care providers could use the Family Stress Checklist to systemically identify a subgroup of adolescent mothers, whose pattern of health care utilization and propensity for mistreating their infants and toddlers suggests the need for this intensive type of support after delivery. Although the consensus is that teenage mothers are at higher psychosocial risk for child abuse and neglect than their adult counterparts (Flanagan et al., 1995; Flanagan et al., 1995; Stevens–Simon & Reichert, 1994; Stier et al., 1993; Zuravin, 1988), like Murphy and colleagues (Murphy et al., 1985) who studied lower socioeconomic status adult mothers, we found that the high risk subgroup, defined by a score of 25 or more on the Family Stress Checklist, represented 40% of the CAMP population. Adolescent mothers who were identified during the prenatal period as being at high risk for mistreating their children were as compliant with health maintenance visits and immunizations, and did not visit the Emergency Department more frequently than their lower risk peers, but they were far more likely to bring their infants to the CAMP clinic for illnesses and other acute medical concerns. It seems unlikely that this disparity was a direct reflection of the health status of the two groups of infants at birth, since the incidence of prematurity and intensive care unit admissions was almost identical in the high and low risk groups. Moreover, even though the high risk infants weighed less at birth, the frequency of acute care visits was unrelated to birth weight or gestational age. Because social factors such as strained maternal relationships and disorganized, overcrowded living environments are known to adversely effect the health of infants (Harris et al., 1989; Leventhal, Pew, Berg, & Garber, 1996), it is possible that the high risk infants actually became less resilient and more susceptible to infection during their first year of life. However, because the high risk group exhibited the pattern of increased utilization almost immediately, and the results of a prior

C. Stevens–Simon et al. / Child Abuse & Neglect 6 (2001) 737–751

747

study suggest that this type of health care utilization is indicative of a stressful mother-infant relationship (Harris et al., 1989), we surmise that the high risk teen mothers had more trouble interpreting their infants’ signals of discomfort and soothing them and, as a result, came in earlier in the course of illnesses or for less severe symptoms (McCarthy, Freudigman, Cicchetti, Mayes, Benitez, Salloum, Baron, Fink, Anderson, & LaCamera, 2000). The observational nature of this study precludes conclusions about causality. Nevertheless, it is interesting to consider how increased continuity of care might be causally related to Emergency Department utilization. Within this context, our finding that these at-risk teenage mothers turned to the CAMP clinic, rather than the Emergency Department, suggests the potential importance of having a medical home base (Sia, 1992). However, the didactic and supportive clinical services provided in CAMP do not appear to have been intensive enough. By the end of the first postpartum year, the high risk mothers had made nearly twice as many acute medical care visits as their low risk counterparts. The magnitude of this difference suggests that an additional benefit of interventions, targeted at preventing the maltreatment of children born to adolescent mothers, might be decreased utilization of costly acute health care services. Within this context, our finding that 60% of CAMP patients made at least one visit to the Emergency Department during the first postpartum year raises concern about the efficacy of telephone triage in this population, and adds to the growing evidence that there is a need for counseling specifically designed to prevent the overuse of acute medical care services in poor, Medicaid dependent populations like this one (American Academy of Pediatrics, 1998; Baker, Schubert, Kirwan, Lenkauskas, & Spaeth, 1999; Barber, King, Monroe, & Nichols, 2000; McCarthy et al., 2000; Sia, 1992). During the first postpartum year, 11% of the infants who were identified as being at high risk for abuse and neglect were mistreated and removed from their mother’s custody, and by the end of the second postpartum year the incidence of maltreatment in this group had risen to 22%. In a substantial proportion of cases, the major change in primary caretaking was not precipitated by abuse or neglect, but by the mother leaving the home and abdicating her child-care responsibilities to a friend or relative. The frequency with which these separations occur in adolescent-headed families (Flanagan et al., 1995; Stier et al., 1993) emphasizes the importance of including counseling, specifically designed to prevent abandonment, in future interventions with socially at-risk adolescent-headed families. After controlling for pre-existing sociodemographic differences, high risk 1-year-olds were 8.41 (95% CI:1.77– 40.01) times and high risk 2-year-olds 5.19 (95% CI:1.99 –13.60) times more likely to have been mistreated than their low risk counterparts. Although all types of maltreatment occurred more commonly in the high risk group, our finding that the majority of the high risk infants were not mistreated at all is consistent with the results of most other studies (Brayden et al., 1993; Flanagan et al., 1995; Leventhal et al., 1989; Murphy et al., 1985; Stevens–Simon & Reichert, 1994; Zuravin, 1988). Collectively, these data show that the sensitivity and negative predictive value of predictions of maltreatment exceeds the specificity and positive predictive values (Brayden et al., 1993; Flanagan et al., 1995; Leventhal et al., 1989; Murphy et al., 1985; Stevens–Simon & Reichert, 1994; Zuravin, 1988). Unlike Flanagan, Coll, Andreozzi, and Riggs (1995), who found that simply not living with a related adult accounted for most of the variance in maltreatment among the participants in their adolescent-oriented parenting program, and Zuravin (1988), who found

748

C. Stevens–Simon et al. / Child Abuse & Neglect 6 (2001) 737–751

that a rapid repeat conception and low educational achievement were the most important mediators of the relationship between adolescent parenthood and child abuse and neglect, we did not find that any single sociodemographic or behavioral characteristic was a significant independent predictor of maltreatment. Rather, it was the clustering of factors, both in present and past life events, that seemed to predispose the adolescents we studied to the dysfunctional parenting styles that typically antedate the maltreatment of infants and children. It seems unlikely that detection bias, resulting from closer follow-up of high risk infants, falsely elevated the rate of abuse in that group (Brayden et al., 1993; Olds, Henderson, Kitzman, & Cole, 1995), given that high and low risk infants made an equivalent number of health maintenance visits. Although there was no formal child abuse prevention program in the CAMP clinic when this study took place, teenagers were referred to supportive services as specific problems arose. It is difficult to know if and how the interventions provided to the high risk families, especially those who made multiple acute care visits to the CAMP clinic, might have effected the incidence of maltreatment in that group. In summary, we have demonstrated that prenatal health care and social service providers can use the Family Stress Checklist to systematically identify a subgroup of adolescent mothers, whose excessive use of the acute care services offered in a specialized adolescentoriented maternity program, and tendency to mistreat their children suggests the need for even more intensive support services after delivery than those offered in the CAMP clinic. Recent research findings suggest that adding a home visitation component to clinic-based programs like CAMP might be a cost effective way to prevent both the overuse of pediatric acute care services and the maltreatment of children in this easily identifiable high-risk cohort of adolescent mothers (American Academy of Pediatrics, 1998; Leventhal et al., 1996; Olds et al., 1986; 1995; 1997; US Advisory Board on Child Abuse & Neglect, 1991).

Acknowledgments The authors wish to thank Grace Cisneros, the staff, and the patients of the Colorado Adolescent Maternity Program for their participation in this study and their help with data collection.

References American Academy of Pediatrics. (1998). The role of home-visitation programs in improving health outcomes for children and families. Pediatrics, 101, 486 – 489. Baker, R. C., Schubert, C. J., Kirwan, K. A., Lenkauskas, S. M., & Spaeth, J. T. (1999). After-hours telephone triage and advice in private and non-private pediatric populations. Archives of Pediatrics & Adolescent Medicine, 153, 292–296. Barber, J. W., King, W. D., Monroe, K., & Nichols, M. H. (2000). Evaluation of emergency department referrals by telephone triage. Pediatrics, 105, 819 – 821. Barnet, B., Joffe, A., Duggan, A., Wilson, M., & Repke, J. (1996). Depressive symptoms, stress, and social support in pregnant and postpartum adolescents. Archives of Pediatrics & Adolescent Medicine, 150, 64 – 69.

C. Stevens–Simon et al. / Child Abuse & Neglect 6 (2001) 737–751

749

Brayden, R., Altemeier, W., Dietrich, M., Tucker, D., Christensen, M., McLaughlin, J., & Sherrod, K. (1993). A prospective study of secondary prevention of child maltreatment. Journal of Pediatrics, 122, 511–516. Flanagan, P., Coll, C. G., Andreozzi, L., & Riggs, S. (1995). Predicting maltreatment of children of teenage mothers. Archives of Pediatrics and Adolescent Medicine, 149, 451– 455. Flanagan, P., McGrath, M., Meyer, E., & Garcia Coll, C. (1995). Adolescent development and transitions to motherhood. Pediatrics, 96, 273–277. Hardy, J., & Street, R. (1989). Family support and parenting education in the home: an effective extension of clinic-based preventive health care services for poor children. Journal of Pediatrics, 115, 927–931. Harris, E. S., Weston, D. R., & Lieberman, A. F. (1989). Quality of mother-infant attachment and pediatric health care use. Pediatrics, 84, 248 –254. Kempe, A., Dempsey, C., Whitefield, J., Botherner, J., MacKenzie, T., & Poole, S. (2000). Appropriateness of urgent referrals by nurses at a hospital-based pediatric call center. Archives of Pediatrics and Adolescent Medicine, 154, 355–360. Kessner, D. M., Singer, J., & Kalk, C. E. (1973). Infant death: an analysis by maternal risk and health care. In Contrasts in health status, Vol. 1 (p. 59). Washington, DC: National Academy of Sciences. Leventhal, J. M., Garber, R. B., & Brady, C. A. (1989). Identification during the postpartum period of infants who are at high risk of child maltreatment. Journal of Pediatrics, 114, 481– 487. Leventhal, J. M., Pew, C., Berg, A. T., & Garber, R. B. (1996). Use of health services by children who were identified during the postpartum period as being at high risk of child abuse and neglect. Pediatrics, 97, 331–335. McCarthy, P., Freudigman, K., Cicchetti, D., Mayes, L., Benitez, J. L., Salloum, S., Baron, M., Fink, H., Anderson, R., & LaCamera, R. (2000). The mother-child interaction and clinical judgment during acute pediatric illness. Journal of Pediatrics, 136, 809 – 817. Murphy, S., Orkow, B., & Nicola, R. M. (1985). Prenatal prediction of child abuse and neglect: a prospective study. Child Abuse & Neglect, 9, 225–235. Newton, R. W., Webster, P. A. C., Binu, P. S., Maskrey, N., & Phillips, A. B. (1979). Psychosocial stress in pregnancy and its relation to the onset of premature labour. British Medical Journal, 2, 411– 413. Nie, N. H., Hull, C. H., & Jenkins, J. G. (1989). Statistical package for the social sciences (3rd ed.). New York: McGraw–Hill: SPSSX. Olds, D. L., Eckenrode, J., Henderson, C. R., Kitzman, H., Powers, J., Cole, R., Sidora, K., Morris, P., Pettitt, L. M., & Luckey, D. (1997). Long-term effects of home visitation on maternal life course and child abuse and neglect: 15-year follow-up of a randomized trial. Journal American Medical Association, 278, 637– 643. Olds, D. L., Henderson, C. R., Chamberlin, R., & Tatelbaum, R. (1986). Preventing child abuse and neglect: a randomized trial of nurse home visitation. Pediatrics, 78, 65–78. Olds, D., Henderson, C., Kitzman, H., & Cole, R. (1995). Effects of prenatal and infancy nurse home visitation on surveillance of child maltreatment. Pediatrics, 95, 365–372. Olds, D. L., & Kitzman, H. (1990). Can home visitation improve the health of women and children at environmental risk? Pediatrics, 86, 108 –116. Radloff, L. (1977). The CES-D scale: a self-report depression scale for research in the general population. Journal Applied Psychologic Measures, 1, 385– 401. Sia, C. C. J. (1992). The medical home base: pediatric practice and child advocacy in the 1990s. Pediatrics, 90, 419 – 423. Smilkstein, G., Ashworth, C., & Montano, D. (1982). Validity and reliability of the Family APGAR as a test of family function. Journal of Family Practice, 15, 309 –311. Stevens–Simon, C., Kelly, L., & Kulick, R. (2000). A village would be nice but it takes a long-acting contraceptive to prevent repeat adolescent pregnancies. Pediatric Research, 47, 13A. Stevens–Simon, C., & Lowy, R. (1995). Is teenage childbearing an adaptive strategy for the socioeconomically disadvantaged or a strategy for adapting to socioeconomic disadvantage? Archives of Pediatrics and Adolescent Medicine, 149, 912–915. Stevens–Simon, C., & McAnarney, E. R. (1995). Change in depressive symptoms during pregnancy: relationship to birth outcome. Journal of Pediatrics and Adolescent Gynecology, 8, 29 –33.

750

C. Stevens–Simon et al. / Child Abuse & Neglect 6 (2001) 737–751

Stevens–Simon, C., & Reichert, S. (1994). Child sexual abuse and adolescent pregnancy. Archives of Pediatrics and Adolescent Medicine, 148, 23–27. Stevens–Simon, C., & White, M. (1991). Adolescent pregnancy. Pediatric Annals, 20, 322–331. Stier, D. M., Leventhal, J. M., Berg, A. T., Johnson, L., & Mezger, J. (1993). Are children born to young mothers at increased risk of maltreatment? Pediatrics, 91, 642– 648. US Advisory Board on Child Abuse and Neglect (1991). Creating caring communities. Blueprint for an effective federal policy on child abuse and neglect. Washington, DC: US Department of Health and Human Services, Administration for Children and Families. US Department of Health and Human Services, Public Health Service (1992). Standards for pediatric immunization practices. Washington DC: US Department of Health and Human Services. Wissow, L. S. (1995). Child abuse and neglect. New England of Journal Medicine, 332, 1425–1431. Zuravin, S. J. (1988). Child maltreatment and teenage first births: a relationship mediated by chronic sociodemographic stress? American Journal of Orthopsychiatry, 58, 91–103.

Re´sume´ Objectif: De´terminer si le Family Stress Checklist sert aux intervenants en services pre´nataux a` identifier les adolescentes qui seraient porte´es a` maltraiter leurs enfants. Me´thode: Les auteurs ont e´tudie´ 262 adolescentes qui faisaient partie d’un programme de pre´paration a` la maternite´. Durant la pe´riode pre´natale, on a administre´ le Family Stress Checklist afin de de´terminer la probabilite´ de mauvais traitements; un score de ⱖ 25 fut conside´re´ le seuil de risque e´leve´. Au moyen d’un questionnaire auto-administre´, on a recueilli des renseignements sur le contexte social de la grossesse et la qualite´ des services de sante´ fournis et on a consulte´ les dossiers me´dicaux. On a classe´ trois types d’interruption des soins maternels, en ordre descendant: les mauvais traitements, la ne´gligence et l’abandon. Re´sultats: Les scores tire´s du Family Stress Checklist variaient de 0 a` 65 (moyenne ⫾ SD ⫽ 20,1 ⫾ 1,4); 113 (43%) adolescentes fut cote´es a` risque e´leve´. Tant les me`res a` risque e´leve´ que celles a` risque infe´rieur ont visite´ des services de sante´ ou des services d’urgence. Cependant, le groupe a` risque e´leve´ a cherche´ de l’aide pour des situations plus graves (6,0 ⫾ 4.1 compare´ a` 3,9 ⫾ 3,3; p ⬍ .0001). Apre`s avoir controˆle´ les diffe´rences socio-de´mographiques pre´existantes, on a note´ que les be´be´s d’un an e´taient 8,41 fois (95% CI: 1,77– 40,01) plus aptes a` avoir connu des traitements abusifs que les be´be´s a` risque infe´rieur. De meˆme en e´tait-il pour les enfants de deux ans, soit 5,19 fois (95% CI: 1,99 –13,60). Conclusions: Les intervenants en services pre´nataux peuvent utiliser le Family Stress Checklist pour identifier de fac¸on syste´matique les me`res adolescentes qui font appel de fac¸on excessive aux services de soins actifs et qui ont tendance a` maltraiter leurs enfants, ces caracte´ristiques portant a` croire qu’elles ont besoin d’appuis spe´ciaux.

Resumen Objetivo: Determinar si el “Family Stress Chechlist” (Inventario de Stress Familiar) ayuda a los que ofrecen cuidados prenatales para identificar las adolescentes que esta´n en riesgo de maltratar a sus hijos. Me´todos: Estudiamos 262 participantes en un programa integral de orientacio´n hacia la maternidad para adolescentes. Durante el periodo prenatal se utilizo´ el “Family Stress Checklist” para cuantificar el potencial de abuso, con los puntajes ⬎ 25 determinando alto riesgo. Se obtuvo informacio´n sobre

C. Stevens–Simon et al. / Child Abuse & Neglect 6 (2001) 737–751

751

el contexto social del embarazo y el patro´n de cuidado de salud utilizado con un cuestionario de autoreporte y revisando el historial me´dico. Los trastornos mayores del cuidado primario que ofrecı´an las madres adolescentes fue clasificado jera´rquicamente como abuso, negligencia y abandono. Resultados: Los puntajes del “Family Stress Checklist” variaron de 0 a 65 (media ⫹ SD ⫽ 20.1 ⫹ 1.4); 113(43%) de los 262 adolescentes fueron clasificados como de alto riesgo. Las madres adolescentes con riesgos Altos y Bajos hicieron el mismo nu´mero de visitas pero, el grupo de alto riesgo inicio´ significativamente ma´s visitas de cuidado agudas (6.0 ⫹ 4.1 comparadas con 3.9 ⫹ 3.3; p ⬍ .0001). Despue´s de controlar las diferencias sociodemogra´ficas pre-existentes, los nin˜os de 1 an˜o de alto riesgo estaban 8.41 (95% CI:1.77– 40.01)veces y los nin˜os de alto riesgo de 2 an˜os 5.19 (95% CI:1.99 –13.60)veces ma´s propensos a haber sido maltratados que sus semejantes de bajo riesgo. Conclusiones: Los que ofrecen cuidados prenatales pueden utilizar el “Family Stress Checklist” para identificar sistema´ticamente un subgrupo de madres adolescentes cuyo excesivo uso de los servicios de cuidado me´dico agudo y tendencia a maltratar a sus hijos sugiere que necesitan servicios de apoyo adicionales.