A randomized controlled trial testing the efficacy of a Nurse Home Visiting Program for Pregnant Adolescents

To test the efficacy of a nurse home visiting program (HVP) on child development, maternal and environmental outcomes in the first years of life. We conducted a randomized controlled trial to test the efficacy of Primeiros Laços, a nurse HVP for adolescent mothers living in a poor urban area of São Paulo, Brazil. Eighty adolescent mothers were included and randomized to receive either Primeiros Laços (intervention group, n = 40) or healthcare as usual (control group, n = 40). Primeiros Laços is a home visiting intervention delivered by trained nurses that starts during the first 16 weeks of pregnancy and continues to the child’s age of 24 months. Participants were assessed by blind interviewers at 8–16 weeks of pregnancy (baseline), 30 weeks of pregnancy, and 3, 6, 12, and 24 months of child’s age. We assessed oscillatory power in the mid-range alpha frequency via electroencephalography when the children were aged 6 months. Child development was measured by the Bayley Scales of Infant Development Third Edition (BSID-III). Weight and length were measured by trained professionals and anthropometric indexes were calculated. The home environment and maternal interaction with the child was measured by the Home Observation and Measurement of the Environment. Generalized estimating equation models were used to examine intervention effects on the trajectories of outcomes. Standardized effect sizes (Cohen’s d) were calculated using marginal means from endpoint assessments of all outcomes. The trial was registered at clinicaltrial.gov: NCT02807818. Our analyses showed significant positive effects of the intervention on child expressive language development (coefficient = 0.89, 95% CI [0.18, 1.61], p = 0.014), maternal emotional/verbal responsivity (coefficient = 0.97, 95% CI [0.37, 1.58], p = 0.002), and opportunities for variety in daily stimulation (coefficient = 0.37, 95% CI [0.09, 0.66], p = 0.009). Standardized effect sizes of the intervention were small to moderate. Primeiros Laços is a promising intervention to promote child development and to improve the home environment of low-income adolescent mothers. However, considering the limitations of our study, future studies should be conducted to assess Primeiros Laços potential to benefit this population. Clinical Trial Registration: The study was registered at clinicaltrial.gov (Registration date: 21/06/2016 and Registration number: NCT02807818).

improvements in child development markers in the first 2 years of life: (1) oscillatory alpha power, an electroencephalography (EEG) measure of brain development, (2) cognitive, language, and motor developmental milestones measured by the Bayley Scales of Infant Development Third Edition (BSID-III), and (3) child growth measured by anthropometric indices. We also hypothesized our intervention would positively impact (4) the quality of stimulation and support available to the child in the home environment measured by the Home Observation for Measurement of the Environment (HOME). Therefore, the objective of the present study is to test the efficacy of Primeiros Laços on child development, maternal and environmental outcomes in the first 2 years of life.

Methods
Design. We conducted a parallel-group randomized controlled trial to test the efficacy of a nurse HVP for adolescent mothers living in a poor urban area in Sao Paulo, Brazil.
Participants. From June to September 2015, eighty low-income pregnant youth were randomized into two groups: intervention (HVP by nurses, n = 40) and control (healthcare as usual, n = 40). Inclusion criteria were: (a) first pregnancy, (b) pregnant youth aged 14-19 years old, (c) low socioeconomic status (classes C, D, E according to the ABEP scale) 45 , (d) pregnancy between 8 and 16 weeks, (e) living in the western region of Sao Paulo. Randomization was stratified by the following characteristics: primary healthcare unit type, and grandmother years of schooling. The allocation ratio used was 1:1. An on-site research manager hired specifically for this study was responsible for randomizing and allocating participants.
Setting. The western region of Sao Paulo city is characterized by high rates of violence, widespread slums, and adverse living conditions. In 2015, the population of the western region was 395 thousand inhabitants with approximately 20.0% of adolescents. Child mortality rate was 5.7-12.4 per 1000 live births and neonatal mortality was 1.7-2.9 per 1000 live births, depending on the district. At the time, the homicide rate ranged from 3.7 to 8.6 per 100 thousand inhabitants 46 . We established a facility within the community at Liga Solidaria's Educandário Dom Duarte (ligasolidaria.org.br), a well-known institution with decades of experience in educational and social services, where our research staff coordinated the study and conducted blinded assessments.
Procedures. Participants were identified in Primary Healthcare Units in the western region of Sao Paulo.
Every potential participant received a brochure with a brief explanation of the study inviting them for an interview with a research assistant. All participants who met inclusion criteria and were interested in the study signed a consent form approved by the institutional review boards of the University of Sao Paulo Medical School (ref: 052/15) and the São Paulo Municipal Health Department.
Participants were assessed by blinded interviewers at 8-16 weeks of pregnancy (baseline), 30 weeks of pregnancy, and 3, 6, 12, and 24 months of infant's age. All interviewers were experienced psychologists who underwent a one-month training provided by study coordinators. Prior to the baseline assessment, interviewers pilottested the protocol assessing voluntary participants living in the Western region.

Sample characterization measures.
Sample characterization at baseline consisted of information on maternal age, ethnicity, school enrollment, educational level, occupation, grandmother's education level, family enrollment in social welfare program, number of people living in the residence, and total family income per month.
Maternal mental health was measured using the Beck Depression Inventory (BDI) 47,48 , the Beck Anxiety Scale (BAI) 49 , and the Adult Attention Deficit Hyperactivity Disorder (ADHD) Self-report Scale (ASRS) 50 . Those are well-known scales validated in Brazil showing adequate psychometric properties [51][52][53] . To characterize the sample at baseline, we used the following cut-offs for each domain/scale: depression (BDI score > 20), anxiety (BAI score > 16), and ADHD (ASRS score > 9). Also, substance, alcohol, and tobacco use (lifetime and during gestation) was measured by a structured questionnaire developed for this study.
The Childhood Trauma Questionnaire (CTQ) 54 was administered to participants to collect information on maternal history of abuse. The CTQ screens for experiences of abuse and neglect during childhood, resulting in five domains: emotional, physical, sexual, emotional neglect, and physical neglect. It has been translated and validated in Brazil 55,56 .
Family food insecurity was measured by the brief version of the Brazilian Food Insecurity Scale (Escala de Insegurança Alimentar, EBIA) 57 , that is based on a widely used scale in Brazil developed to measure food insecurity at the national level with the objective of providing reliable data for federal programs aimed at eliminating hunger 58 . Intervention. Primeiros Laços is a home visiting intervention delivered by trained nurses based on three theoretical frameworks: attachment theory 59 , self-efficacy theory 60 , and the bioecological model 61 . Primeiros Laços was based on the Brazilian program Janelas de Oportunidades 62,63 , the Minding the Baby program 64 , and the NFP 36 , with input from key national stakeholders involved in early childhood and maternal health research and advocacy. The program targets first-time pregnant adolescents and their child, starting during the first 16 weeks of pregnancy and continuing until the child's age of 24 months. It has a learning-based approach aimed at helping the mother prioritize the child in her life in order to improve maternal sensitivity to the child's behaviors and emotions. Nurses visited participants according to the following plan: (a) biweekly during gestation and from 2 to 20 months of child's age, (b) weekly during the first and last month of pregnancy, and the puerperium, and (c) monthly from 21 to 24 months of child's age. www.nature.com/scientificreports/ The Primeiros Laços content is directed to five domains: (1) health and social care: promotion of maternal and child health focused on nutrition, hygiene, common pathologies in childhood, domestic care, vaccination, prevention of accidents, and early child development; (2) environmental health: support to ensure mothers have adequate living conditions, safe housing, access to health services, and daycare and school for children; (3) life course: support to help adolescent mothers plan future goals, such as finishing high school, finding a part-time job, finding a career, planning to go to college, and postpone the birth of a second child. Life-course planning was discussed individually according to personal goals and objectives; (4) parenting skills: education on child development and parenting skills with the objective of helping the adolescent mother to develop healthy mother-child interactions in order to ensure a secure attachment style; (5) family and social support: the role of family members and friends to support the needs of mother-child dyad was highlighted. Family members living with the adolescent mother or in the community were also invited to home visits to be part of the intervention process. Nurses were supervised weekly by senior nurses (A.C., L.F.) and a child psychologist. More details about the intervention can be found elsewhere 64-67 . Care as usual. Participants allocated to the control group received healthcare from the Unified Health System (Sistema Único de Saúde, SUS), Brazil's public health system 68 , according to national guidelines [68][69][70][71] that are in line with the World Health Organization guidelines. The primary care system delivers prenatal and postnatal care with an emphasis on the prevention of gestational problems. Cases of gestational risk were referred to specialized health services.
Outcome measures. We assessed oscillatory power in the infant alpha frequency (6-9 Hz) during a quasiresting-state condition (while infants watched videos of abstract shapes) using EEG when the children were aged 6 months. The data were recorded using a 128-channel Geodesic Sensor Net and a NetAmp 200 DC-coupled amplifier (Electrical Geodesics Inc., Eugene, OR, USA). Absolute power in the alpha band (6-9 Hz) was computed for each electrode and subsequently averaged across clusters of electrodes in the frontal, central, righthemisphere posterior, left-hemisphere posterior and occipital scalp regions, following previous work 72 . Alpha power in each cluster was used in analysis. EEG data collection was attempted with 50 children at age 6 months. Of these, 19 were excluded due to technical problems with the EEG system that resulted in corrupted data files (n = 4), poor-quality recordings (n = 10) or fewer than 20 artefact-free epochs (n = 5).  73 administered by a trained psychologist at 3, 6, 12, and 24 months of the child's age. The BSID-III consists of a series of play tasks and behavior observations and usually takes between 45 and 60 min to administer. The BSID-III has been translated to Brazilian Portuguese and validated [73][74][75][76] , showing good psychometric properties. We assessed motor, language, and cognitive development. Since there is no Brazilian normative data available for the BSID-III raw scores were used in all analyses.
Child anthropometric growth was measured at birth, 6, 12, and 24 months of the child's age. We measured child weight and length/height using standard procedures. Children's BMI-for-age and length-for-age were calculated based on WHO growth standards 77 via WHO Anthro version 3.2.2 SPSS syntax 78 . Z-scores ≤ 4 and ≥ 4 were excluded because those were considered measurement errors (BMI-for-age of 1 participant time-point from intervention, length-for-age of 1 participant time-point from control).
The quality of stimulation and support available to the child in the home environment was assessed via the Home Observation for Measurement of the Environment (HOME) 79,80 . HOME is a 45-item instrument aimed at assessing the caring environment where the child is being reared. Information is obtained via observation of the home environment and interview questions with the primary caregiver. The scores can be divided into six domains: emotional and verbal responsivity, avoidance of restriction and punishment, organization of the physical/temporal environment, provision of appropriate play materials, parental involvement with the child, and opportunities for variety in daily stimulation. It was translated to Brazilian Portuguese and it has been applied in different regions and contexts in the country 80-83 . Sample size. The sample size was calculated based on the difference in EEG alpha frequency power between the groups (30%) with a probability of type I error of 5% and statistical power of 80%.
Statistical analysis. EEG data were analysed using one ANOVA model with the between-subjects factor group (intervention vs. control) and the within-subjects factor electrode cluster (frontal, central, right-hemisphere posterior, left-hemisphere posterior, occipital) with alpha power as the dependent variable. A significant interaction between group and electrode cluster was further investigated with Bonferroni-corrected planned pairwise contrasts between the groups at each electrode cluster.
Generalized estimating equation (GEE) 84 models were used to examine intervention effects on the outcomes of interest that were measured longitudinally (child development and anthropometric growth, and quality of stimulation and support available to the child in the home environment). We used quasi-likelihood under the independence model criterion (QIC) to detect covariance structures that were optimal for unequally spaced observations for each model. Time was entered as a continuous covariate in all models. Fitted models were used to estimate and plot marginal mean scores for continuous outcomes at each time point. Time trends were verified by checking if a quadratic term was statistically significant 85,86 . We used a modified intention-to-treat approach in our analysis: all randomized participants were included in the analyses, except participants with missing data at  Intervention effects on child anthropometric development are depicted in Fig. 4 (BMI-for-age: intervention n = 36, control n = 39; Length-for-age: intervention n = 35, control n = 37). GEE models did not show statistically significant effects of the intervention on BMI-for-age and length-for-age (Fig. 3). Table S3 depicts length-for-age and BMI-for-age categorized by z-score. At 24 months of age, the majority of the sample was in the normal range of length-for-age (86.4%) and BMI-for-age (57.1%). Figure 5 shows the intervention effects on the mother-child relationship and the home environment measured by the HOME scale (intervention n = 30, control n = 32). Maternal emotional and verbal responsivity (coefficient = 0.  Table S1 reports the number of participants included in the analyses for each outcome. Only participants without data in at least one time point were not included in the analyses (e.g., stillbirth, neonatal death, etc.). Differences at baseline between participants included vs. not included in the efficacy analyses were tested (Table S2). The majority of participants enrolled in school at the time of the baseline assessment provided information on outcomes of the study. Maternal age (p = 0.004), mother enrolled in school (p = 0.005), and maternal history of physical abuse (p = 0.012) were associated with inclusion in the EEG outcome analysis. Mother enrolled in school (p = 0.004) and maternal history of emotional abuse (p = 0.045) characteristics were associated with being included in the child development, measured by the BSID-III, analysis. Being enrolled in school (p = 0.019) was associated with inclusion in the HOME data analysis. Finally, the following characteristics were associated with inclusion in the anthropometric growth analysis: mother enrolled in school (p = 0.016 for BMI-for-age, p = 0.008 for length-for-age), maternal history of emotional abuse (p = 0.016 for BMI-for-age, p = 0.028 for length-for-age).

Discussion
Our study showed positive effects of a nurse HVP, the Primeiros Laços program, for impoverished adolescent mothers living in adverse conditions. The intervention group showed statistically significant benefits on child expressive language, as well as on two domains representing the quality of the home environment: maternal emotional/verbal responsibility and opportunities for variety in daily stimulation. To our knowledge, this is the first study testing an HVP in Brazil using an RCT design. The impact of the intervention on child expressive language development replicates findings from one of the studies that tested the efficacy of NFP 90 . In this study, children of mothers who received visits from nurses were www.nature.com/scientificreports/ www.nature.com/scientificreports/ less likely to exhibit language delays at 21 months and had a higher development score measured by the BSID at 24 months compared with children in the control group 90 . Given that the NFP and the Primeiros Laços program share several components, it is not surprising that part of our results is in line with NFP findings 90 . The observed effects of the intervention on child expressive language development could be explained by the impact of the intervention on the variety of daily stimulation and maternal emotional/verbal responsivity. More variety in daily stimulation increases the chances of expanding the child's general behavior repertoire. Moreover, it is well-known that, aside from influences of developmental changes in perception, cognition, and phonology, the development of children's communication and language skills are supported by the way the caregivers respond to the child's interactions [90][91][92][93] . In addition to that, longitudinal studies have demonstrated an association between variations in child language development and maternal responsiveness 93 .  www.nature.com/scientificreports/ The effect of Primeiros Laços on opportunities for variety in daily stimulation and maternal emotional/verbal responsivity may be explained by the improved parenting skills component of the intervention. Nurse home visits after children were born emphasized child development and early cognitive and socioemotional stimulation. Nurses taught participants the importance of early stimulation, specifically showing activities that can help to enhance the mother-child bond and to provide adequate stimulation for the child's age. In line with these findings and intervention components, a previous study analyzing the present sample showed the Primeiros Laços program increased the probability of early stimulation behaviors by mothers at 18 months of the child's age 94 .
Even though we found the Primeiros Laços program had a statistically significant effect on child language development and home environment outcomes, these effects were small to moderate according to reported Cohen's d (0.37-0.55). This may be attributed to the fact that the Primeiros Laços program has intervention components divided into five domains (health and social care, environmental health, life course, parenting skills, family, and social support), not all exclusively related to child development and/or maternal caregiving skills. Moreover, participants were adolescent mothers living in adverse conditions, meaning that some problems related to food insecurity, housing, violence, among others, are prioritized in home visits, leaving less time to be dedicated to parenting skills in some cases. This may explain the small to moderate effect size on child expressive language development. Also, maternal responsivity is a complex and nuanced construct that may require more attention, time, and specific strategies to have a more significant impact on child language.
No effect of the intervention was found for oscillatory alpha power, suggesting that the home-visiting program did not result in positive effects on brain maturation by the time the children were aged 6 months. While these findings contrast with our hypotheses and previous work in children exposed to psychosocial deprivation 72,95 the null findings are not particularly surprising given that we could not collect EEG data from the entire sample or at all of the age time-points, as we had initially planned, due to the technical difficulties with the EEG system. It may be the case that differences in oscillatory activity associated with the home-visiting program emerge later in development than age 6 months.
We did not find an effect of the Primeiros Laços program on child anthropometric development. This finding could be explained by political changes that influenced health indicators in the last decades. The inception of the Unified Health System (Sistema Único de Saúde) in 1988, a public health system with universal coverage and a focus on primary health care programs 68 led to the drastic decline in the prevalence of stunting 96 and low weight-for-age 97 . Data suggests that virtually every Brazilian pregnant woman have access to prenatal care (97.8%) 98 . Considering this context, it may be difficult to implement interventions that can add to the strong effects of such programs.
Our findings should be viewed in light of some limitations. First, our sample size is small, which precluded analyses to understand the effects of the intervention in specific subgroups (for example, a comparison of intervention effects in participants with higher versus lower exposure to food insecurity, housing, violence, etc.). Moreover, the small sample size precluded a mediation analysis to better understand whether there is a connection between increased maternal emotional/verbal responsivity and opportunities for variety in daily stimulation, and the expressive language improvements observed in children from the intervention group. Second, our study was powered based on the potential effect of the intervention in EEG alpha wave frequency, so our findings should be interpreted with caution, since there is a possibility that we did not find an effect of the  www.nature.com/scientificreports/ intervention in some of the outcomes because of our sample size. Third, our assessment of the home environment, stimulation, and mother-child relationship was based exclusively on the frequency of behaviors measured by the HOME scale. We did not assess the quality of the interaction between participants and their children using a more fine-grained measure. For instance, maternal responsiveness is composed of only 11 items with binary response options (yes/no), therefore lacking a full range of maternal behaviors and dimensional detail. Likewise, we did not assess the involvement of the child's father or other father figures in the home 99 . Fourth, by the time children completed 24-months of age, 40% of the intervention group were not receiving home visits anymore and only 67.5% received at least 50% of planned home visits. The dropout rate and limited number of visits delivered can potentially explain the lack of findings on some outcomes, or the small to moderate effects found on developmental and environmental outcomes. Fifth, the randomization and allocation procedures were not conducted by an external researcher, potentially contributing to a bias in this process. However, we did not find any evidence of balancement issues favoring the intervention. Sixth, we were not able to collect oscillatory alpha power data according to our original plan due to EEG technical difficulties beyond our control. This issue significantly limited the available data for our EEG analysis, limiting our statistical power and biasing our results. Finally, baseline characteristics were associated with participants being included in our outcome analysis (e.g., maternal age, mother enrolled in school, maternal childhood trauma). This may have biased our overall findings. These characteristics should be considered in future studies similar to ours. Targeting these characteristics via intervention mechanisms could help prevent loss to follow-up. Lastly, the long-term outcomes of our intervention remain to be determined. Future assessments of the study population will allow us to investigate whether the effects on child language development are maintained. If indeed they are, they will add to the extant literature showing the potential of HVP programs to prevent and promote child health, as rates of language impairments in the first years of life are estimated to be around 65% in children with low socio-economic background 100 . The possibility of preventing such impairments in these populations represents an enormous gain for the community, as language delay is associated with impairments in reading ability 101 , school readiness 102 , and persistent communication problems, which have been shown to negatively affect children's development of prosocial behaviors (i.e., behaviors involving empathetic actions, as well as putting the needs of others first, and having a positive sense of social responsibility, and which are very important for children to build better social relationships when entering formal school) [102][103][104][105] . Follow-up studies will also be important to observe whether differences between intervention and control groups appear in other cognitive domains. In NFP, superior executive functioning and behavioral adaptation during testing was observed in a follow-up conducted two years after the program ended (at age 4) (Olds et al., 2004). Finally, from a public health perspective, it will be important to document in future long-term outcome studies whether or not the mothers who participated in Primeiros Laços went on to give birth to fewer children and if there was greater spacing between first and second pregnancies for those mothers who did have additional children 106,107 .

Conclusions
Our findings demonstrate the efficacy of Primeiros Laços, a nurse HVP for pregnant adolescents living in an urban deprived area, on child expressive language, as well as on two domains representing the quality of the home environment: maternal emotional/verbal responsibility and opportunities for variety in daily stimulation. To our knowledge, our study is the first to use a randomized controlled trial design to test the efficacy of an HVP intervention on child and home environment outcomes in Brazil. Primeiros Laços is a promising intervention able to promote the development of children born of low-income adolescent mothers. Due to its structured and manualized nature, Primeiros Laços could be implemented in the context of the public health system, therefore working alongside well-established programs. Since adolescent pregnancy is a prevalent public health issue in Brazil and other LMICs, an evidence-based intervention such as Primeiros Laços is timely. However, considering the limitations of our study (e.g., small sample size, dropout rate) future studies should be conducted to assess Primeiros Laços potential to benefit adolescent mothers and their children. www.nature.com/scientificreports/