Urban families are faced with a multitude of psychosocial problems such as poverty, single parent homes, domestic violence, inadequate role models for parenting, and lack of information about available resources. Such an environment is not conducive to good parenting, parent-child interactions, or child development. Even the most comprehensive health care cannot solve all psychosocial problems, but care which enhances social support and provides additional parenting education may help minimize some of them. Home visitors and parenting programs may provide social support and increase parental knowledge, but are time-consuming and expensive.¹

The traditional model of individual well child care (IWCC) is not necessarily the best format for providing social support or information about child development to all types of parents. Group well child care (GWCC), where the provider facilitates discussion of child-rearing issues with a group of parents of similarly aged children, is an attractive alternative. GWCC offers increased time for parent education and prolonged opportunities for the provider to observe parent-child interactions as well as extended time for providing appropriate role-modeling and advice.⁴ In addition, GWCC has been shown to improve delivery of health care information on topics such as safety, nutrition, behavior, and infant sleep.⁵ In middle-class populations the GWCC format was found to allow for significantly more discussion of personal issues and daily infant care than during individual visits.⁶

Maternal-child interactions and child development are enhanced when pediatricians focus on the child's development during well child care visits.⁷ GWCC, by maximizing discussion of child-rearing issues, offers the potential for increasing positive maternal-child interaction and optimizing development, without the additional costs of home visitation or parenting programs. In middle-class, white families, GWCC has been associated with increased maternal knowledge of child care and development as compared with families whose children received IWCC.⁸

Previous research on GWCC has concentrated on middle-class families, with outcome measures focusing on increases in parental knowledge.^4,8 Theoretically, GWCC might be most efficacious for disadvantaged families by providing a structured social support network.^9,10 Numerous studies have demonstrated the positive effects of social support on mother-child attachment and maternal attitudes and behaviors, especially among the most vulnerable populations.¹¹ We postulated that, in high-risk families, GWCC would result in improved parenting and improved child development compared with IWCC, without consuming more resources.

METHODS

Families were enrolled in the study when the child was approximately 3 months old. At enrollment, children were randomized to receive either IWCC or GWCC. Randomization was carried out separately at each clinic, with assignment to GWCC or IWCC balanced for each block of 10 enrollees. At enrollment, mothers of study infants completed several baseline questionnaires including the Clinic Intake Questionnaire, the Social Support Questionnaire, the Difficult Life Circumstances scale, and the Parenting Stress Index. The Clinic Intake Questionnaire covers the child's health history and demographic information about the family, and also includes screening questions to assess maternal substance abuse, depression, and history of abuse as a child.¹⁵ The screening questions on substance abuse, depression, and history of abuse have previously been validated by comparisons with other, widely used, instruments.¹⁵ The Social Support Questionnaire measures the adequacy of social support by assessing the number of individuals who provide support to a person in times of stress, and the satisfaction with this support.¹⁹ The Difficult Life Circumstance assesses chronic stress.²⁰ It was chosen over other instruments that measure stressful life events because it focuses on common family problems. The Parenting Stress Index assesses parental sense of competency, and feelings of isolation as a parent.²¹ Normative data are available, and scores correlate with parenting problems such as child abuse and later scores on the child's adjustment.²² The responses on these questionnaires constituted each study child's baseline characteristics.

Study health supervision visits were scheduled at 4, 5, 6, 8, 10, 12, and 15 months of age. The timing and frequency of study visits coincided with the schedule for well child care visits in other programs for high risk families at clinic 2. All care was provided by 2 study nurse practitioners. Prior to beginning the project, a curriculum of topics to be covered at each health supervision visit was developed. The curriculum was directed towards general parenting topics such as nutrition, development, and anticipatory guidance. While all of these topics were ultimately aimed at improving maternal-child interaction and development, the curriculum was not designed to improve performance on any specific outcome measure. The curriculum was followed by the nurse practitioners at both IWCC and GWCC visits.

Children randomized to receive IWCC had traditional one-to-one health supervision visits provided by the study nurse practitioners. Those randomized to GWCC were assigned to a cohort of infants with birthdays within 2 months of each other. GWCC health supervision visits consisted of a group discussion of child rearing issues led by the nurse practitioners, lasting 45 to 60 minutes. A brief physical examination of children receiving GWCC either preceded or followed each group session. Immunizations and health screening were provided to all study children regardless of assignment to GWCC or IWCC. If a child failed to keep a study visit, systematic attempts were made to schedule another health supervision appointment. At these rescheduled appointments, the curriculum for the age appropriate health supervision visit was followed, but all care was delivered using the individual well child care format regardless of the patient's assignment to GWCC or IWCC.

After the last study visit (at approximately 15 months of age), public health nurses visited the homes of study infants and completed two scales of maternal-infant interaction: the Nursing Child Assessment Teaching Scale (NCATS), and the Home Observation for Measurement of the Environment (HOME). The public health nurses were trained and experienced in these assessments, and were blind to the randomization status of study children.

The NCATS consists of 73 items and six scales that assess maternal-infant interaction including: sensitivity to cues, response to infant distress, social-emotional growth fostering, clarity of cues, and responsiveness of the child to the parent.²³ Global scores of the NCATS at 12 to 24 months of age are strongly predictive of intellectual and language outcomes at 4 years of age as measured by the Stanford-Binet.²⁴

The HOME is designed to detect environments likely to be suboptimal for the cognitive development of children at high risk.²⁵ Scores on the HOME correlate with later psychologic development and school performance.^26,27 Because of the young age of the study population, the preschool version was used. This version of the HOME includes a toy inventory and six subscales: emotional and verbal responsitivity of the mother, avoidance of restriction and punishment, organization of the physical and temporal environments, provision of appropriate play materials, and opportunities for variety in daily stimulation.

In addition to the home visits, development of study infants was assessed with the Bayley Scales of Infant Development (Bayley). Two standardized scores were obtained from the Bayley: a Mental Development Index and a Psychomotor Development Index. Scores on the Bayley have a significant correlation with Stanford-Binet scores later in life.²⁸ Bayley evaluations were performed when the children were between the ages of 14.5 and 18 months old. The evaluations were performed by an experienced psychometrician who is trained in the administration of the Bayley and who was blinded to the infants' randomization to GWCC or IWCC.

Intensive, systematic attempts were made to contact each enrolled family for Bayley, HOME, and NCATS evaluations as the study child reached 15 months of age. These efforts continued until 100 of each of these outcomes were obtained in each of 50 GWCC and IWCC children. For the HOME and NCATS, the public health nurse made at least four attempts to schedule a home visit; if this was unsuccessful, the child was excluded from the evaluation. Likewise, four attempts were made to schedule the Bayley assessment for each study patient. For children who failed to show for an evaluation, the Bayley was rescheduled. No further efforts were made for patients who failed to attend this rescheduled appointment. Baseline characteristics of children who completed at least one of the three outcomes were compared to eligible children who failed to complete any outcome evaluations to assess potential dropout bias.

Data on any child who completed an outcome were analyzed, according to assignment to GWCC or IWCC, regardless of whether he or she actually came to any (or all) study visits (ie, intention to treat analysis). Standardized scores for the mental and psychomotor portion of the Bayley in children receiving GWCC were compared to scores in IWCC patients by the use of t-tests. Scores on the HOME and NCATS were dichotomized as high risk or not high risk as defined by the developers of the measures (scores on HOME <32, and <51 on NCATS classified as high risk).^23,25 The proportion of high-risk scores for each of these measures in children in the GWCC group was compared to the proportion of high-risk scores in those receiving IWCC using ² or Fisher's exact test as appropriate. Differences in mean scores for the HOME and NCATS tests between children in the GWCC and IWCC groups were assessed with t-tests.

To control for potential confounding, logistic regression was used to compute risk estimates for high risk scores on the HOME and NCATS, and linear regression was used to assess the association between mental and motor score on the Bayley and assignment to GWCC or IWCC. We analyzed 16 covariates (baseline characteristics) as potential confounders. In each separate analysis, a variable was considered to be a confounder and was retained in the final regression model if adjustment for it altered the crude relative risk (for logistic analyses) or  score (linear analyses) by 10% or more. Differences in the relationship between covariate status and the specific outcomes resulted in each outcome measure being adjusted for by a slightly different set of covariates. For all statistical tests, differences were considered significant when the two-sided P value was <.05.

Based on a sample size of 50 children in each group, the study had a 90% chance of detecting a 10 point difference between mean scores of patients in the IWCC and GWCC groups on either the mental or psychomotor portion of the Bayley (two-sided  level = .05, with IWCC mean scores assumed to be 100, and a standard deviation of 15 points). In a study of a local ongoing parenting education program, 19% of the children had high-risk scores on the NCATS, and 35% had high-risk HOME scores.²⁹ Assuming these proportions of high-risk scores among IWCC children, the study had an 80% chance of detecting a decrease to 3% high-risk scores on the NCATS, and a 80% chance for detecting a decrease to 10% high-risk scores on the HOME assessment among patients receiving GWCC (two-tailed  level = .05).

In addition to Bayley, HOME, and NCATS evaluations, patient compliance and provider time were also measured. Compliance with well child care appointments was measured two ways. Study compliance was defined as the number of scheduled study visits attended. Overall compliance was measured by totaling the number of any health supervision visit (either study or rescheduled) during the appropriate time period. Differences in study and overall compliance between children in the GWCC and IWCC groups were assessed with t-tests. Provider time was measured for all study visits. For IWCC patients, provider time was defined as the number of minutes spent in the examination room by the nurse practitioner. For children receiving GWCC, provider time was calculated by adding the per child group session time (total minutes of the group session/number of children attending the session) and time spent during the physical examination of the child. Differences in provider time were evaluated using generalized estimating equations.³⁰

The study was approved by the Institutional Review Board of Children's Hospital and Medical Center. Signed informed consent was obtained from the parents of all study patients.

RESULTS

Baseline characteristics of children who completed at least one outcome and characteristics of those who, although eligible, did not have any outcome measures obtained are summarized in Table 1. Among those randomized to GWCC, baseline characteristics of those who completed an outcome were similar to those in whom no outcomes were obtained. There were minor differences between completers and noncompleters among those randomized to IWCC; specifically, mothers of children in whom an outcome was obtained were significantly less likely to report being neglected as a child compared to those whose children did not have any outcome measures (P = .03). When children who completed an outcome were compared to those who did not, regardless of randomization to IWCC or GWCC, there were no significant differences for any baseline characteristic.

Table 1. Comparison of Baseline Characteristics of Children Who Completed at Least One Outcome to Those Who Did Not Complete an Outcome by Assignment to GWCC or IWCC [View Table]

The 114 patients who completed a developmental outcome came from high-risk families. Overall, 26% of the mothers of these infants were less than 20 years old at the time of delivery, 65% were unmarried, 31% had less than a high school education, and 46% reported a household income of less than $500 per month. In addition to the 41% with a positive screen for alcohol abuse, and the 13% with a positive screen for drug abuse, 36% of mothers of study patients had a positive screen for depression, 14% were physically abused, and 16% sexually abused as children, and 61% reported that their social support was less than satisfactory.

Univariate analysis of the three study outcomes is presented in Table 2. The mean standardized scores on the mental and psychomotor portions of the Bayley evaluation were similar in GWCC and IWCC patients. After adjusting for potentially confounding variables there was no significant association between randomization to GWCC or IWCC and either the mental index (P = .97) or the psychomotor index (P = .21). Maternal-child interaction, as measured by the NCATS, was virtually identical in children receiving GWCC and those in the IWCC group. Both mean scores, and proportion of high risk scores, on the NCATS were similar regardless of assignment to group or individual well child care. After controlling for social support and ethnicity of the child, the odds ratio (OR) for IWCC as a predictor of a high risk maternal-child interaction was 0.51 (95% confidence interval, 0.10, 2.50). Finally, although mean scores on the HOME were similar in children receiving GWCC and those in the IWCC group, 16% of IWCC patients and 4% of GWCC children had scores indicative of a high-risk home environment (P = .096). After adjusting for maternal education and a history of drug abuse, children assigned to IWCC were over four times more likely than GWCC patients to have a high-risk home environment, as measured by the HOME (OR = 4.9); however, the confidence interval was 0.65 to 37.4, indicating that the OR was not statistically significantly different than 1.0. 

Table 2. Comparison of Developmental Outcomes Between Children Receiving GWCC and IWCC [View Table]

Children in the IWCC group attended significantly more of the seven scheduled visits than those receiving GWCC (mean number of study visits: 4.61 ± 1.94 and 3.51 ± 2.32, respectively, P = .007). However, the overall compliance with any health supervision (including make-up visits) was similar in the two groups; the mean number of age-appropriate well child care visits was 5.93 ± 2.10 for patients in the GWCC group versus 5.86 ± 1.99 for those receiving IWCC (P = .86). Provider time was also similar (mean number of minutes per study visit: 19.8 ± 5.6 for GWCC patients and 20.4 ± 6.7 for those receiving IWCC, P = .66).

DISCUSSION

There are several possible explanations for why outcomes among children in the GWCC group were not superior to those receiving IWCC. The most obvious is that GWCC may not be effective in promoting better development and/or maternal-child interaction than IWCC. Previous research has found that home visitation programs may improve development in some high-risk children.¹ However, these programs are costly. We postulated that GWCC, with its increased focus on child development, and by fostering peer social support, would provide some of the advantages of home visitation programs, but in a typical clinical setting and without increased costs. However, the intervention may not have been intense enough to have a significant impact. During an eleven month period each child had seven visits scheduled; parents of children in the GWCC group attended, on average, only half of these scheduled sessions. Most successful home visitation programs have included 10 or more visits.¹

Because not all eligible children had outcome measures, it is conceivable that the results may be biased. Presumably those infants at lowest risk would be the most likely to complete the study protocol, regardless of assignment to GWCC or IWCC. This would tend to minimize any differences in the measured effectiveness of the two methods for delivering well child care. However, the baseline characteristics of the mothers of children in whom an outcome measure was obtained were similar to characteristics of mothers of children who did not have any outcome evaluations, suggesting that dropout bias was minimal.

Our study may have had insufficient power to detect clinically significant differences in outcomes between GWCC and IWCC assigned children. Prior to beginning the project, sample size calculations were based on the assumption that the percentages of high-risk scores on the HOME and NCATS evaluations among children assigned to IWCC would be similar to those of children enrolled in a previous study at clinic 2.²⁹ In that project, 33% of parents who completed the program had high-risk scores on the HOME and 19% had high-risk NCATS scores. In contrast, in our current study, the proportion of children receiving IWCC who had high-risk scores on these evaluations were 16% and 10%, respectively. Although these differences in outcomes may be due to differences in the populations studied, it is more likely that participation in the study led to improved outcomes, regardless of assignment to GWCC or IWCC. Parents of children in both groups received extraordinary attention from the study nurse practitioners and social workers, including telephone calls, and birthday cards. The resulting improved outcomes in both groups would tend to bias the results towards the null hypothesis.

We purposely chose outcome measures that have been shown to predict the future level of functioning in children. HOME scores are correlated with school performance,²⁷ NCATS with language development at 4 years of age,²⁴ and Bayley results with Stanford-Binet scores.²⁸ Despite this approach, it is possible that longer follow-up is needed to detect differences between the GWCC and IWCC groups. Achenbach et al reviewed results of a intervention designed to optimize caretaking interactions between mothers and their low birth weight infants.³¹ Mothers in the treatment group received 11 one-on-one sessions aimed at facilitating adaptation to their low birth weight infants. There were no significant differences between the intervention and control groups at 6, 12, and 24 months of age. However, at 3 years of age, those infants in the intervention group had significantly higher scores on the McCarthy General Cognitive Index than control children (P < .05). The higher scores among those in the intervention group continued until the end of the follow-up period at 9 years of age. This study suggests that the positive effects of GWCC may only become apparent after an extended follow-up period.

This study highlights both the problems and potential of GWCC for high-risk children. The lack of flexibility in scheduling group sessions resulted in a significantly lower show rate for patients assigned to GWCC when compared to traditional individual care. It was gratifying that the parents of GWCC children had enough positive feelings about the care that they received to reschedule missed appointments, with overall compliance being similar in both groups. We have shown the GWCC does not require additional provider time. Our data suggest that high-risk children do not need one-on-one well child care in order to have good outcomes. Longer term follow-up will be needed to ascertain if the tendency towards better outcomes among GWCC recipients will persist and/or become significantly better than those who received IWCC.

ACKNOWLEDGMENTS