Implementing Family Connects Meets HHS Criteria

Last updated: October 2014

This report summarizes information on how a given model was implemented in the studies reviewed. The report includes only information provided in (1) implementation studies and (2) effectiveness studies that rate moderate or high. These studies vary in the level of detail they provide about implementation features. Thus, the report does not provide an exhaustive picture of how the model was implemented across the programs studied. HomVEE notes, in the text or in parentheses, the number of studies that reported information on a given implementation feature.

Implementation experiences

Summary of sources

Information in this section is based on studies included in the HomVEE review. For Family Connects, we reviewed two randomized controlled trials (RCTs) and two standalone implementation studies. (Please see Studies for Implementation Experiences for a list of the studies and to link to the characteristics of the samples examined in the effectiveness studies.)

All studies were implemented under the program name Durham Connects, but we use the name Family Connects to reflect the name used in program sites as of June 2014. One implementation study was conducted early in the model’s implementation, as developers were refining the model; therefore, subsequent studies might report slightly different program information. The other three studies are based on the same randomized trial. Two of them examined the same subsample of participants; as a result of this overlap, we refer to these as one study throughout the rest of this section.

View Revisions

Characteristics of model participants

Family Connects was offered to all postpartum mothers residing in Durham County, North Carolina (all studies). Approximately 2,325 families were eligible to participate (two studies). Among the eligible families for whom data were available (2,279 families), 36 percent of mothers identified as African American; 29 percent identified as white; 22 percent identified as Latina; 5 percent identified as Asian/Asian American; and 8 percent identified as multiracial, other, or unknown. Of the 1,596 families that completed the program, 40 percent of mothers identified as white; 37 percent as African American; 23 percent as multiracial, other, or unknown; and 26 percent also identified as Hispanic (one study). The studies did not include information on the ages of program participants.

Approximately 60 percent of eligible families received Medicaid or had no insurance; nearly half of those who completed the program (44 percent) were married (one study).

Participation in the program was voluntary (all studies).

View Revisions

Location and setting

The program operated in Durham County, North Carolina (all studies). The studies did not characterize Durham County as urban, rural, or suburban.

According to the initial study, the program was implemented through a partnership between researchers associated with the Duke Endowment’s Durham Family Initiative (DFI), and the Durham County Public Health Department. The research team, based at the Center for Child and Family Policy at Duke University and the Center for Child and Family Health (CCFH), designed the program and the evaluation. The Durham County Public Health Department provided nurses, supervisory staff, and some computer support. The health department and the research team at Duke University continued to implement the program, according to a more recent study.

View Revisions

Staffing and supervision

Public health nurses conducted home visits, as required by the model (three studies). One study reported that additional staff included three trained recruiters to visit mothers in the hospital; two clinical nurse supervisors based at CCFH; one supervisor of nurses at the health department; a leadership team composed of DFI researchers and health department personnel; and a community outreach team. The outreach team raised awareness of the program and developed partnerships with local health providers and community agencies. The outreach team was staffed by a director, three outreach workers, a part-time interpreter, and a data manager.

None of the studies reviewed included information on the education of supervisors or other staff. 

The studies reviewed did not include information on preservice training. According to one of the studies, nurses occasionally received in-service training during weekly meetings with nurse supervisors and the leadership team.

The studies did not describe the nature of the nurses’ supervision, the supervisors’ caseloads, or the nurse home visitors’ caseloads.

View Revisions

Model services

Recruiters visited women in hospitals after their deliveries to describe the program, schedule the home visit, and provide incentives for participation (according to the more recent studies). Participants were offered one home visit 3 to 8 weeks postpartum, with up to four additional visits, depending on their needs. In the earlier study, staff initially attempted to conduct the home visit at 4 weeks postpartum and later at 12 weeks postpartum, with the possibility of a follow-up visit if there were serious concerns (such as depression or domestic violence). Nurses attempted to confirm appointments a few days before home visits (two studies).

During the home visits, nurses examined the babies; conducted risk assessment interviews; educated parents on topics specific to their assessed needs (such as lactation or crying); and provided support, advice, and referrals to community support services (three studies). The risk assessment interview focused on 12 risk factors across four domains: parental and infant health care, infant care giving, family violence and safety, and parental mental health and well-being (according to the more recent study). Families determined to be at imminent risk were offered an immediate intervention (the study did not describe the intervention). Nurses sent summaries of the visit to the families’ pediatricians, noting any issues that arose (according to the earlier study). Program staff called or visited families approximately one month after the last visit to learn about new needs and encourage families to follow through with referrals (two studies).

The studies did not discuss whether services were available in languages other than English. One study reported that families were recruited by staff who spoke English and Spanish.

View Revisions

Model adaptations or enhancements

None of the studies described adaptations or enhancements.

View Revisions


One study reported that 1,863 of the eligible families (80 percent) completed at least the first session and 1,596 completed the full program (86 percent of families that scheduled an initial home visit and 69 percent of eligible families). Nurses covered, on average, 14 of 20 lessons with each family. The study also found that 60 percent of families reported connecting with a referred provider and 39 percent reported receiving the referred service. The studies did not include information about the average duration of visits or total number of visits received.

View Revisions

Lessons learned

The studies discussed lessons learned about implementation, including lessons related to family engagement, flexibility, community engagement, and use of data.

Engagement in the model was associated with family resources and risk factors (two studies). Authors in one study analyzed hospital birth records to examine associations between engagement and risk factors. Across ethnic groups, families with high-risk births (low birth weight, birth complications, or low gestational age) were less likely than those with low-risk births to schedule and complete a home visit. This study also reported that families with high demographic risks (defined as being uninsured or having public insurance, neighborhood poverty, or young maternal age) were more likely than low-risk families to schedule a home visit, but were relatively less likely to complete the visit. This finding was consistent across ethnic groups. The authors suggested that home visiting staff should not necessarily interpret missed visits as a lack of model interest. Anecdotal information from home visitors suggested that low-resource families faced a greater number of participation barriers, including greater mobility and disconnected telephones than their high-resource counterparts. The second study learned through staff interviews that most of the refusals came from middle- and upper-income mothers who felt they did not need the service or did not like the idea of a visit from a public health nurse.

Another study asked a subset of 830 participating families about their satisfaction with the model. Nearly everyone considered the model to be a helpful resource for materials that the nurses provided, such as diapers, books, or thermometers (99 percent of families); discussing their needs (98 percent); and education (95 percent). Almost all of the families (99 percent) would recommend the visit to a new mother.

Lessons related to service flexibility, community engagement, and use of model data, according to interviews conducted in one study, included the following:

  • Staff observed that flexibility to conduct home visits after regular business hours would help accommodate families’ schedules. Such hours, however, would be outside the typical work schedule of public health department nurses;
  • The model should fit within the broader array of services already available to residents, and providers in the other service systems (such as medical providers) should be engaged in the model; and
  • model developers highlighted as a strength their ability to inform model decisions with model data. For example, their analysis of data from early phases of research led them to implement Family Connects as a universal home visitation model.
View Revisions