Parenting interventions to promote early child development in the first three years of life: A global systematic review and meta-analysis
Jeong, Franchett, Ramos de Oliveira, Rehmani, Yousafzai · 2021
공개 학술지 PLoS Medicine 에 실린 총설 논문입니다(2021년). CC BY 4.0 으로 공개된 글이고, 그림 · 표 · 참고 문헌 표시를 뺀 본문을 실었습니다.
이 장은 원문만 있습니다. 문장별 해석은 앞 장부터 차례로 붙입니다.
3장 · Parenting interventions to promote early child development in the first three years of life: A global systematic review and meta-analysis — Results
Study selection and inclusion
A total of 11,906 unique records were identified from electronic databases. Fourteen additional records were identified from article references and subject matter expertise. A total of 11,580 records were excluded based on title and abstract screening. The full texts of the remaining 340 articles were reviewed, and 229 articles were excluded for not meeting inclusion criteria. The final set comprised of 111 articles representing 102 unique RCTs (Fig 1).
Sample, intervention, and study characteristics
Table 1 presents the characteristics of the 102 RCTs of parenting interventions. Trials were implemented across a total of 33 countries. The majority of trials were conducted in HICs (61 trials in 14 countries), compared to LMICs (41 trials in 19 countries). The most represented countries were the United States of America (40 trials), Bangladesh (8 trials), Jamaica (6 trials), and India (5 trials). Studies were published between 1974 and 2020. Analytic sample sizes ranged from 32 to 3,202 individuals. The average age of children at baseline varied across interventions: 16 trials (16%) enrolled pregnant mothers, 30 (29%) enrolled children with an average age of 3 months and younger, 24 (24%) enrolled children with an average age of 3 to 12 months, and 32 (31%) enrolled children who were on average older than 12 months. Intervention duration ranged from a brief intervention delivered over 1 month to longer interventions delivered across the first 5 years of the child’s life. Nearly all interventions (93%) focused on supporting mothers exclusively, with only 7 interventions (7%) additionally engaging fathers to some degree.
Out of the total 102 interventions, 70 (69%) had some degree of a responsive caregiving component that aimed to improve parent–child interactions. Of these responsive caregiving interventions, 37 interventions (53%) involved facilitators who observed live parent–child interactions and/or utilized video recordings in order to provide feedback for promoting parental sensitivity and responsiveness in the context of play, communication, reading, and/or feeding. The other 33 responsive caregiving interventions (47%) more generally aimed to improve the quality of parent–child interactions through guidance, demonstrations, and encouragement of parents to practice during sessions and on their own.
The remaining 32 interventions (31%) did not include responsive caregiving components. Of these, 22 (69%) provided general parenting support with information and guidance on child development and parenting topics, and 10 (31%) focused on increasing parental stimulation and/or provision of learning materials. Among the total 102 interventions, 45 (44%) also integrated messages beyond parenting for ECD, such as maternal and child nutrition, child health, caregiver mental health, and early education.
Interventions utilized a range of delivery models: 58 interventions (57%) were delivered to individual caregivers and families, 13 (13%) were exclusively group based, and 31 (30%) incorporated both individually delivered and group-based components. Delivery settings also varied. Nearly half or 47 interventions (46%) were delivered exclusively through home visits, 11 in a clinical or hospital context (11%), 11 in a community space (11%), and 32 interventions (31%) used a combination of multiple settings. One intervention was delivered virtually.
Risk of bias assessment
The total risk of bias score across all studies was moderate (mean = 5.1, SD = 1.5) ranging from 2 to 9 on a scale of 0 to 16. In general, risk of bias was low for random sequence generation, blinding of outcome assessors, and incomplete outcome data (S1 Fig). However, risk of bias for allocation concealment and selective reporting were unclear for the majority of studies. Given the nature of psychoeducational and behavioral parenting interventions that involve parents’ active participation, blinding of participants was not possible. Egger’s tests suggested evidence of small-sample bias for 2 out of the 10 outcomes: child language development, z = 2.74, P = 0.01; and parent–child interactions, z = 3.78, P < 0.001.
Effects on early child development outcomes
Fifty-eight studies provided a total of 58 effect sizes for children’s cognitive development. Approximately half of these studies (56%) reported any details regarding reliability or validity of the measure of cognitive development. The Bayley Scales of Infant and Toddler Development (BSID) was the most commonly used measure in more than half the studies (54%). See S2 Table for full list of measures used for each of the outcomes. The pooled result showed a moderate positive impact of parenting interventions on improving cognitive development (SMD = 0.32, 95% CI: 0.23, 0.40, P < 0.001; I² = 89%, P < 0.001; Fig 2).
Forty-six studies provided a total of 58 effect sizes for children’s language development. Approximately half of these studies (57%) reported any details regarding reliability or validity of the measure of language development. The BSID (25%) and the MacArthur-Bates Communicative Development Inventory (14%) were the most commonly used measures. The pooled result showed a moderate positive impact on improving language development (SMD = 0.28, 95% CI: 0.18, 0.37, P < 0.001; I² = 91%, P < 0.001; Fig 3).
Thirty-five studies provided a total of 45 effect sizes for children’s motor development. Approximately half of these studies (56%) reported any details regarding reliability or validity of the measure of motor development. The BSID was the most commonly used measure in more than half the studies (56%). The pooled result showed a moderate positive impact on improving motor development (SMD = 0.24, 95% CI: 0.15, 0.32, P < 0.001; I² = 84%, P < 0.001; Fig 4).
Twenty-nine studies provided a total of 31 effect sizes for children’s socioemotional development. Nearly 2 in 5 studies (38%) reported any details regarding reliability or validity of the measure of socioemotional development. All measures were caregiver reported and most commonly included the Ages and Stages Questionnaire (16%), Ages and Stages Questionnaire: Social-Emotional (13%), and the Brief Infant–Toddler Social and Emotional Assessment (13%). The pooled result showed a small positive impact on improving socioemotional development (SMD = 0.19, 95% CI: 0.10, 0.28, P < 0.001; I² = 81%, P < 0.001; Fig 5).
Twenty-nine studies provided a total of 35 effect sizes for children’s behavior problems. Nearly half of the studies (47%) reported any details regarding reliability or validity of the measure of behavioral problems. All measures were caregiver reported, and the Child Behavior Checklist was the most commonly used measure in more than half the studies (53%). The pooled result showed a small impact on reducing behavior problems (SMD = −0.13, 95% CI: −0.18, −0.08, P < 0.001; I² = 53%, P < 0.001; Fig 6).
Eleven studies provided a total of 11 effect sizes for infant–caregiver attachment. The majority of studies (85%) reported any details regarding reliability or validity of the measure of attachment. The Ainsworth Strange Situation Procedure was the most commonly used measure in 2 out of 5 studies (38%). The pooled result showed a moderate positive impact on improving attachment (SMD = 0.29, 95% CI: 0.18, 0.40, P < 0.001; I² = 0%, P = 0.92; Fig 7).
Effects on parenting outcomes
Sixty-four studies (63%) measured at least 1 parenting outcome, and 35 studies (34%) measured 2 or more parenting outcomes. Sixteen studies provided a total of 16 effect sizes for parenting knowledge. Less than half of studies (44%) reported any details regarding reliability or validity of the measure of parenting knowledge. Nearly all studies (88%) used a knowledge questionnaire that was developed by the authors. The pooled result showed a large positive impact on improving parenting knowledge (SMD = 0.56, 95% CI: 0.33, 0.79, P < 0.001; I² = 97%, P < 0.001; Fig 8).
Thirty-five studies provided a total of 40 effect sizes for parenting practices. Two-thirds of studies (66%) reported any details regarding reliability or validity of the measure of parenting practices. The Home Observation of Measurement of the Environment Inventory-Infant and Toddler version was the most commonly used measure in more than half of studies (58%). The pooled result showed a moderate positive impact on improving parenting practices (SMD = 0.33, 95% CI: 0.22, 0.44, P < 0.001; I² = 93%, P < 0.001; Fig 9).
Twenty-seven studies provided a total of 27 effect sizes for parent–child interactions. Three in four studies (77%) reported any details regarding reliability or validity of the measure of parent–child interactions. Nearly all studies directly observed and coded quality of mother–child interactions (91%). The pooled result showed a moderate positive impact on improving parent–child interactions (SMD = 0.39, 95% CI: 0.24, 0.53, P < 0.001; I² = 93%, P < 0.001; Fig 10).
Twenty-four studies provided a total of 25 effect sizes for parental depressive symptoms. More than half of studies (59%) reported any details regarding reliability or validity of the measure of depressive symptoms. The Center for Epidemiologic Studies Depression Scale was the most commonly used measure in 2 out of 5 studies (41%). The pooled result did not indicate a significant reduction in caregiver depressive symptoms (SMD = −0.07, 95% CI: −0.16, 0.02, P = 0.08; I² = 76%, P < 0.001; Fig 11).
Moderator analyses
Given the large observed heterogeneity across all but one of the pooled effect size estimates, we conducted several moderator analyses to investigate potential sources of this heterogeneity. Tables 2 and 3 present the effects on ECD and parenting outcomes, respectively, stratified by the 7 characteristics: country-income level, average child age at baseline, intervention content, intervention duration, delivery, setting, and study quality. Parenting interventions had a greater effect on child cognitive, language, and motor development and parenting practices in LMICs compared to HICs, and the subgroup differences for these outcomes were statistically significant (e.g., effects on cognitive development: SMD = 0.41, 95% CI: 0.29, 0.53 in LMICs versus SMD = 0.17, 95% CI: 0.10, 0.22 in HICs; P < 0.001 for difference between subgroups; Table 2). Although effects on child socioemotional development, parenting knowledge, parent−child interactions, and parental depressive symptoms did not significantly differ by country-income level, the magnitudes of the estimates were consistently greater for all outcomes in LMICs versus HICs (Table 3).
Parenting interventions that promoted responsive caregiving had significantly greater effects on child cognitive development, parenting knowledge, practices, and parent–child interactions compared to interventions without responsive caregiving (e.g., effects on parenting practices: SMD = 0.42, 95% CI: 0.29, 0.56 for interventions with responsive caregiving components versus SMD = 0.11, 95% CI: 0.01, 0.22 for those without; P = 0.001 for difference between subgroups; Table 3). With the exceptions of greater subgroup effects on cognitive development for programs targeting children older than 12 months of age and greater effects on parenting practices for programs less than 12 months in duration, there were no other subgroup differences on any outcomes by child age, intervention duration, delivery, setting, or study risk of bias score (Tables 2 and 3).
이 글은 위 출처의 원문을 그 라이선스(CC BY 4.0)에 따라 옮겼습니다. 참고문헌 목록과 인용 번호, 그림 · 표 · 각주 · 수식 · 부록은 빼고 절별로 나누어 실었으며, 그 밖의 문장은 원문 그대로입니다. 원 저작자와 발행 기관은 이 화면을 만들거나 검토하지 않았고, 이곳의 해석과 낱말 풀이는 공식 번역이 아닙니다.
