Integrating a Parenting Intervention With Routine Primary Health Care: A Cluster Randomized Trial
OBJECTIVE: More than 200 million children globally do not attain their developmental potential. We hypothesized that a parent training program could be integrated into primary health center visits and benefit child development.
METHODS: We conducted a cluster randomized trial in the Caribbean (Jamaica, Antigua, and St Lucia). Fifteen centers were randomly assigned to the control (n = 250 mother-child pairs) and 14 to the intervention (n = 251 mother-child pairs) groups. Participants were recruited at the 6- to 8-week child health visit. The intervention used group delivery at 5 routine visits from age 3 to 18 months and comprised short films of child development messages, which were shown in the waiting area; discussion and demonstration led by community health workers; and mothers’ practice of activities. Nurses distributed message cards and a few play materials. Primary outcomes were child cognition, language, and hand-eye coordination and secondary outcomes were caregiver knowledge, practices, maternal depression, and child growth, measured after the 18-month visit.
RESULTS: Eight-five percent of enrolled children were tested (control = 210, intervention = 216). Loss did not differ by group. Multilevel analyses showed significant benefits for cognitive development (3.09 points; 95% confidence interval: 1.31 to 4.87 points; effect size: 0.3 SDs). There were no other child benefits. There was a significant benefit to parenting knowledge (treatment effect: 1.59; 95% confidence interval: 1.01 to 2.17; effect size: 0.4).
CONCLUSIONS: An innovative parenting intervention, requiring no additional clinic staff or mothers’ time, was integrated into health services, with benefits to child cognitive development and parent knowledge. This is a promising strategy that merits further evaluation at scale.
- CDI —
- Communicative Development Inventory
- CHW —
- community health worker
- CI —
- confidence interval
- ECD —
- early child development
- HOME —
- Home Observation for Measurement of the Environment
- IQR —
- interquartile range
- PPVT —
- Peabody Picture Vocabulary Test
- WHO —
- World Health Organization
What’s Known on This Subject:
More than 200 million children <5 years are not reaching their developmental potential. Lack of stimulating caregiving is a major cause, and effective scalable interventions are needed. Integrating parenting with health services has been recommended, but there are few evaluations.
What This Study Adds:
An innovative parenting intervention can be delivered at routine visits for primary health care, with benefits to child cognitive development and parenting knowledge. This approach using films, discussion, and practice has the potential for delivery at scale.
Early childhood is a critical period for brain development, and exposure to risks can modify brain structure and function.1,2 More than 200 million children <5 years in low- and middle-income countries do not attain their developmental potential because of risks associated with poverty.3,4 These children achieve lower levels of education, which is associated with lower adult income.3 Investment in early childhood can reduce societal inequality, with benefits across the life course.5,6 For example, home visits to improve parent-child interactions led to educational, behavioral, and income gains in adulthood.7,8
Many agencies, including the World Bank, the Inter-American Development Bank, the United Nations Children’s Fund, and the World Health Organization (WHO), have recognized child development as a priority9,10 and have called for integrating early child development (ECD) with health and nutrition programs.9,11 Small-scale trials show that integrating ECD with nutrition does not diminish the impact of the individual interventions12; however, there have been few larger evaluations and there is a need for programs that are feasible and effective at scale. We developed a parent training package that could be delivered without additional staff and hypothesized that it could be integrated into routine primary health care visits, with benefits for parents’ knowledge, stimulation provided, and children’s developmental levels. We incorporated videos in the package because this method has the potential to reach more families and can be used to demonstrate new skills.13
Study Design and Participants
The evaluation was a cluster randomized trial conducted in Jamaica, Antigua, and St Lucia with public health center as the unit of randomization. The structure of the primary care health system is similar across the 3 countries, with government health centers that provide free maternal and child health services. Routine services are delivered by nurses assisted by community health workers (CHWs). The English-speaking islands of the Caribbean have comparable levels of human development,14 and countries were selected after the Ministries of Health indicated willingness to participate. Primary care services are available in all parishes/regions, and parishes were selected for logistical reasons. In Jamaica, the parishes of Kingston and St Andrew have a higher urban population than other parishes. Attendance is high at the postnatal clinics because immunizations are given and are required for school entry. Immunization rates are 93% to 99%.14 Primary health centers serve predominantly lower and lower-middle income groups.
In Jamaica, of 24 centers in the parishes of Kingston and St Andrew, 4 were excluded (see Fig 1 for details). Ten of the remaining 20 centers were randomly selected for the study. In St Lucia and Antigua, health centers in northern St Lucia and northern and western Antigua were included. Ten centers in each country were randomly assigned to treatment or control groups by using a computer-generated randomization sequence. Randomization was performed by an independent statistician who was unaware of the identity of the health centers.
Thirty children were enrolled from each center in Jamaica and 10 each from centers in Antigua and St Lucia, where the population is smaller. After selection, 1 center in St Lucia closed and was dropped from the study. Ten additional participants were recruited from 2 other centers in the same arm of the trial.
Mothers and infants were recruited at the 6- to 8-week postnatal clinics. Consecutive mothers were recruited until the numbers per clinic were reached. Infants born preterm, multiple births, those aged ≥10 weeks, or those admitted to the special care nursery for >48 hours after birth were excluded. Participants were excluded if they intended to use a different center for child immunizations or if there was no consistent caregiver (Fig 1). Recruitment was conducted from August 2011 to March 2012.
A Jamaican trial of home visiting using CHWs showed an effect size of 0.8 SDs.15 Because of the lower intensity of the new intervention, we hypothesized an effect size of 0.375 SDs for developmental quotient and an intracluster correlation of 0.03. With 15 health centers per group, a sample comprising 10 children in each health center achieves 80% power to detect the hypothesized effect. The sample enrolled was 250 in the control group and 251 in the intervention group.
Short films were developed in collaboration with media consultants experienced in health education (Development Media International, London, United Kingdom). Topics were taken from our previous home-visit interventions,16,17 selecting parental behaviors considered central to promoting child development. The process involved discussion of topic objectives with the film producer, development and review of scripts, filming, and editing to ensure fidelity with objectives. Filming was conducted in Jamaica with 5 mother-child pairs. Nine modules, each ∼3 minutes in duration, covered the following topics: love, responding and comforting, talking to children, praise, using bath time to play and learn, looking at books, simple toys to make, drawing and games, and puzzles. The films showed mothers doing the behaviors we wished to encourage. DVDs comprising a set of 3 topics repeated twice were produced. A different combination of 3 topics was shown when the children in the evaluation sample attended the health visits at 3, 6, 9, 12, and 18 months of age. This approach allowed topics to be shown on more than one occasion.
In each center, CHWs discussed the activities shown with the mothers and demonstrated them. Viewing of films and discussion were conducted in the waiting area while mothers waited to see the nurse. Mothers practiced the activities with their children and were encouraged to make them part of their daily routine. Median duration of the discussion sessions was 16 minutes, with an interquartile range (IQR) of 14 to 20 minutes, and the median number of mothers during the sessions was 37 (IQR: 26–50).
CHWs had a minimum of 3 years’ secondary-level education and received preservice training of up to 20 weeks or in-service training, with limited information on child development. Training in the intervention comprised 3-day workshops with viewing of films and role play. CHWs were given manuals that provided the steps and content for each health visit. Before a new set of topics was shown, a supervisor visited the clinic, reviewed the topics with the CHWs, and provided guidance in discussions and practice. The supervisor monitored implementation quality every 6 weeks using 3-point ratings of how well the CHW involved the mothers and acknowledged and praised their efforts. The median rating of 8 (maximum: 9; IQR: 6–9) suggests that CHWs conducted the sessions satisfactorily. The supervisor provided supportive feedback after the session.
All children were seen by a nurse at each visit. The nurses gave the mothers message cards that reinforced the topics on the films and reviewed the cards with them. They encouraged the mothers to do the activities and to watch the films if they had not done so. At ages 9 and 12 months, nurses gave the parents a picture book, and at 18 months a 3-piece puzzle to take home. The estimated additional time for the nurses per mother was 2 to 3 minutes. We used materials from our home-visit intervention.17 Book reading encourages maternal-child interaction and is associated with later cognitive and language development.18 We selected 2 simple picture books so mothers could talk about the pictures and hoped that mothers would be encouraged to continue with other books. Puzzles build several skills related to the perception of shapes as well as task orientation, persistence, and attention. We selected a puzzle that helped mothers teach concepts such as in/out and big/little and the shapes circle and square.
Centers in the control group provided usual care. All mothers attending intervention centers also received the parenting intervention whether or not they were in the evaluation. The study was explained to mothers invited to participate in the evaluation and written informed consent was obtained. It was explained that a new parenting program was being implemented in some health centers and that the center they attended would either continue as usual or receive the new program.
Ethical approval was obtained from the University of the West Indies Ethics Committee and from the Ministry of Health Advisory Panel on Ethics and Medico-Legal Affairs, Jamaica; the South East Regional Health Authority, Jamaica; the Ministry of Health, Antigua; and the St Lucia Medical and Dental Council.
Mothers were interviewed in the health center to obtain information on the following: mothers’ age, education, and occupation; whether the child’s father lived in the home; crowding (persons per room); and household possessions. Mothers’ vocabulary was measured with the Peabody Picture Vocabulary Test (PPVT),19 and maternal depressive symptoms with the Center for Epidemiologic Studies–Depression Scale.20 Both measures have been piloted and used previously in Jamaica.
A parenting knowledge scale was developed with items from earlier questionnaires used in Jamaica.15 Statements covered topics such as age of introduction of learning activities and importance of maternal-child interaction (Appendix), and mothers indicated whether they agreed completely, agreed a little bit, disagreed a little bit, or disagreed completely.
Information on the child’s date of birth and birth weight was obtained from health records or, when not available, maternal recall. Child length, weight, and head circumference were measured by using standard procedures. Interviewers were trained and interobserver reliability determined comparing each of 5 interviewers with the trainer. Intraclass correlation coefficients were ≥0.94 for length measurements, 0.92 for head circumference, and equal to 1 for weight (a digital scale was used) (n = 10–17 children per interviewer). Coefficients were ≥0.99 for the PPVT, 1 for the parenting knowledge score, and ≥0.98 for the depression scale (n = 15 per interviewer).
Measurements were conducted between February and September 2013. Measurements were conducted in a room at the health centers a minimum of 2 weeks after the 18-month clinic visit. Primary outcomes were children’s development and vocabulary measured by using the cognitive, language, and eye and hand coordination subscales of the Griffith Mental Development scales21 and the MacArthur-Bates Short Form of the Communicative Development Inventory (CDI).22 The Griffiths scales have been modified for Jamaica and are predictive of future IQ and school achievement.23 The developmental quotient is calculated from the subscales and we used the UK test norms. The CDI was piloted and some words were changed to others more familiar to Caribbean children (eg, “bear” to “goat”). Interobserver reliabilities between the trainer and testers for both tests were ≥0.98 (n = 9–13 per tester).
Secondary outcomes included child growth, maternal depressive symptoms, and parenting knowledge. Two additional items were added to the knowledge scale. Maternal practices were assessed with 4 subscales from the Home Observation for Measurement of the Environment (HOME;24 Involvement, Responsivity, Acceptance, and Learning Materials), administered at the clinic by interview and observation of mother and child. Interobserver reliability for the HOME was ≥0.90 (n = 12–16 per interviewer). Interviewers and testers were blind to center assignment. Measurements were not conducted on child health clinic days, because these days were when the intervention was done.
Analyses were conducted by using multilevel models with health center at level 2 and child at level 1. Random effects were specified at the health center level. Differences in baseline characteristics were assessed in the total enrolled sample and those assessed at follow-up. Intervention impacts were estimated including fixed effects for a constant term, group assignment, child’s gender and age, country, and the following baseline variables: birth weight, height-for-age z score, adolescent mother (age ≤19 years), mother's PPVT score, mother's depressive symptoms, mother's educational grade, household crowding and possessions, and the baseline value of the outcome variable if collected. We included these covariates because child development was not measured at baseline (age 6 weeks). Hence, covariates that are plausibly associated with development might improve the precision of the estimates.
Inverse probability weighting was used to correct for possible bias because of loss to follow-up.25,26 A logistic regression for the variable “assessed at trial end” (yes = 1, no = 0) was estimated over a constant term and adolescent mother status (the only variable associated with loss). The probability of being assessed was calculated and inverse probability weights computed.
Analyses were by intention to treat with intervention effects tested on 5 primary and 6 secondary outcomes. Multiple-comparison procedures were used to correct the risk of type 1 error (false positive). P values for the null hypothesis were adjusted by using the Holm-Šidák step-down procedure.27,28 The family-wise type 1 error rate was fixed at 0.05 across the 11 outcomes tested.
Statistical analysis was carried out with Stata version 13.1 (StataCorp, College Station, TX).29 z Scores for height-for-age, weight-for-height, and head circumference for age were calculated by using the WHO growth standards (WHO Anthro version 3.2.2; World Health Organization, Geneva, Switzerland)
We enrolled 501 mother-child pairs, and 426 were tested at the end of the trial. The children’s mean age on assessment was 19.7 months (SD: 0.9 months). The trial profile with reasons for loss to follow-up is given in Fig 1. There were no significant differences between groups in enrollment characteristics (Table 1). Loss did not differ by group (16% of children were lost to follow-up in the control and 14% were lost to follow-up in the intervention). Adolescent mothers were more likely to be lost to study follow-up than other women (P = .013). There were no other significant differences between participants tested and those lost to study follow-up. The follow-up sample had a greater proportion of boys in the intervention group (P = .021). No other characteristics differed by treatment group (Table 1).
Attendance was high in both groups. The children’s mother usually brought their child to the clinic, with 97.9% of children accompanied by their mother at the 3-month visit declining to 89.4% at 18 months; 83.1% of mothers attended all 5 visits. When mothers did not attend, children were brought by their father, other relatives, and occasionally by a nonrelative. According to parental report, <1% of children missed any of the visits.
The intervention had a significant benefit to children’s cognitive development, with a treatment effect of 3.09 points (95% confidence interval [CI]: 1.31 to 4.87), which is equivalent to an effect size of 0.3 SDs (Table 2). Without adjustment for covariates or loss to follow-up, the treatment effect was similar (3.12 points; 95% CI: 0.86 to 5.39; Holm-Šidák P value = .07). There were no benefits to the language or hand and eye subscales, or the CDI vocabulary score.
Mothers in the intervention group improved significantly more in parenting scores than the control group (Table 2) (treatment effect: 1.59; 95% CI: 1.01 to 2.17; effect size: 0.4). There were no effects of intervention on the other secondary outcomes. The children’s mean height-for-age, weight-for-height, and head circumference z scores were close to zero, indicating that growth was comparable to the WHO growth standards.
A parenting intervention implemented during routine primary health care visits in 3 Caribbean countries improved child cognitive development and mothers’ knowledge of child development. The size of the cognitive benefit was comparable to other more intensive programs elsewhere,30 although it was smaller than previous Jamaican home-visiting interventions.15,17
There were no benefits to the children’s language or fine-motor development. In the current intervention, there was only 1 contact after 12 months, which is a time of rapid expansion in language. Moreover, the children’s mean language scores were age appropriate and it may be easier to benefit developmental domains where deficits are greatest.
There are few evaluations of group delivery of ECD interventions for children in this age range. Weekly group sessions on responsive feeding and stimulation in Bangladesh benefited children’s language,31 and a trial in Pakistan reported moderate effect sizes from monthly home visits and group sessions conducted from ages 3 to 24 months.32 The children in those studies were considerably more disadvantaged than in the current study and had greater deficits in development.
The benefit to parenting knowledge indicates that mothers remembered the messages shown. The mothers in the films were of a social background similar to the women attending the clinics, which may have helped them see the behaviors depicted as relevant. Video modeling can assist with learning new health behaviors.13,33 In addition, the discussion, demonstration, and practice would have reinforced the messages. Videos could be beneficial as part of parenting interventions with the use of other group-delivery platforms such as community mothers’ groups. Benefits of showing videos alone should be investigated because there is some evidence that videos improve parent knowledge of other health behaviors.34,35
There were no intervention benefits for the HOME scores, which may require more modeling and practice than could be provided in the limited number of contacts in busy clinics. Maternal depressive symptoms were also not reduced, although benefits were seen from a home-visit parenting program.36 It is likely that the relationship between mother and visitor is an important component of benefits for depression.
There were several challenges to implementation. Clinics were often noisy and crowded, and some mothers would have had difficulty hearing and seeing the films. It also made it difficult for the health workers to interact with the mothers during the demonstration.
Linking the intervention to primary care in these Caribbean countries has the advantage of excellent coverage and compliance with child health visits. The intervention required investment in equipment, materials, and staff training, but it was implemented by the existing clinic staff. It also did not require additional time at the clinic for the parents. There were no adverse effects of adding the intervention on nutrition or immunization status of the children, and other small studies suggest that ECD and nutrition activities can be integrated without negative impacts on either component.12 However, this issue will need attention, especially in countries in which children’s health and nutrition are poorer and the capacity of health services more limited.
The intervention was supervised by the research team, which may have facilitated success.37 In planning for scale-up there is a need to identify staff to provide supervision. Furthermore, health service contacts are less frequent after 18 months and other programs will be required.
Our results show that it is feasible to integrate a parenting intervention in child health clinics without the need for additional staff or asking mothers to spend extra time at the clinic. Despite low intensity and challenging clinic conditions, the intervention benefited child cognitive development and parent knowledge. This innovative approach merits further evaluation to determine if similar benefits can be achieved at scale.
APPENDIX: PARENT KNOWLEDGE SCALE
Too much love and attention will spoil a child.
A parent needs to spank or beat young children when they are rude or they will grow up to be bad.
It is important that a busy mother spend plenty of time talking with her infant.
It is important that parents look at picture books with children who are <2 years old.
The best way to get a child to behave is to praise him/her when he/she is good.
It is important that a busy mother spend plenty of time playing with her young child.
There is no need to give toys to children <1 year old.
A time for play is important for young children.
Singing and chatting with your infant will help him/her learn.
Children should not be given crayons until they are ready to learn to write.
Young children should not be held when they cry because this will make them want to be held all the time.
How a parent behaves with her child when she/he is young affects how well she/he will learn in school.
Parents were asked to indicate if they agreed completely, agreed a little bit, disagreed a little bit, or disagreed completely with each statement. Items 1, 2, 7, 10, and 11 were reverse coded. Items 11 and 12 were added at follow-up assessment.
We thank the Ministry of Health, Jamaica, in particular, the Kingston and St Andrew Health Department, and the Ministries of Health of Antigua and St Lucia for facilitating the study and their cooperation throughout. We thank the intervention supervisors and the evaluation team for their commitment and professionalism. We also thank the nurses and community health workers for their contribution to the success of the study and to the parents who made the study possible.
- Accepted May 8, 2015.
- Address correspondence to Susan P. Walker, PhD, Tropical Medicine Research Institute, The University of the West Indies, Kingston 7, Jamaica. E-mail:
Drs Walker, Grantham-McGregor, Chang, and Powell conceptualized and designed the study with input from Dr Lopez-Boo and Dr Baker-Henningham; Dr Chang coordinated the implementation of the study and oversaw training of the evaluation team; Drs Powell, Chang, and Walker led the development of the films, training manuals, and intervention materials with input from Drs Grantham-McGregor and Baker-Henningham; Dr Powell led the training of the intervention team (supervisors, nurses, community health workers) with input from Drs Baker-Henningham and Chang; Dr Vera-Hernández analyzed the data in consultation with Drs Walker and Grantham-McGregor; Dr Walker drafted the manuscript with contributions from Drs Chang and Vera-Hernández; and all authors contributed to the critical review and revision of the manuscript and approved the final manuscript as submitted.
This trial has been registered with the ISRCTN Register (identifier ISRCTN43108304).
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: Supported by the Inter-American Development Bank.
POTENTIAL CONFLICT OF INTEREST: Dr Lopez-Boo is employed by the Inter-American Development Bank; the other authors have indicated they have no financial relationships relevant to this article to disclose.
- Heckman JJ
- Campbell F,
- Conti G,
- Heckman JJ,
- et al
- Gertler P,
- Heckman J,
- Pinto R,
- et al
- Walker SP,
- Chang SM,
- Vera-Hernández M,
- Grantham-McGregor S
- ↵United Nations Development Programme. Human Development Report 2014. New York, NY: United Nations Development Programme; 2014
- Powell C,
- Baker-Henningham H,
- Walker S,
- Gernay J,
- Grantham-McGregor S
- Dunn LM,
- Dunn DM
- Radloff L
- Griffiths R,
- Huntley M
- Fenson L,
- Marchman VA,
- Thal DJ,
- Dale PS,
- Reznick S,
- Bates E
- Caldwell BM,
- Bradley RH
- Wooldridge J
- Holm S
- ↵StataCorp. Stata statistical software. Release 13 [computer program]. College Station, TX: StataCorp LP; 2013
- Hamadani JD,
- Huda SN,
- Khatun F,
- Grantham-McGregor SM
- Yousafzai AK,
- Rasheed MA,
- Rizvi A,
- Armstrong R,
- Bhutta ZA
- Baker-Henningham H,
- Powell C,
- Walker S,
- Grantham-McGregor S
- Copyright © 2015 by the American Academy of Pediatrics