Programs that Work
Healthy Families New York (HFNY)
Healthy and Safe Children
Children not experiencing physical, psychological or emotional abuse
Babies born weighing more than 5.5 pounds and improving outcomes for low birth weight babies
Age of Child
Early Childhood (0-8)
Type of Setting
Type of Service
Health Care Services
Type of Outcome Addressed
Child Abuse and Neglect
Substance Use and Dependence
Evidence Level (What does this mean?)
Based on the national Healthy Families America (HFA) critical program elements, Healthy Families New York (HFNY) is a community-based prevention program that seeks to improve the health and well-being of children at risk for abuse and neglect by providing intensive home visitation services. The target population consists of expectant parents and parents with an infant less than three months old who are considered to be at high risk for child abuse and neglect. Specially trained paraprofessionals are assigned to the participating families to deliver home visitation services until the child reaches five or is enrolled in Head Start or kindergarten. Home visitors provide families with support, education, and referrals to community services aimed at addressing the following goals: 1) promoting positive parenting skills and parent-child interaction; 2) preventing child abuse and neglect; 3) ensuring optimal prenatal care and child health and development; and 4) increasing parents' self-sufficiency. HFNY began operation in 1995 as an initiative of the New York State Office of Children and Family Services in collaboration with the New York State Department of Health, Prevent Child Abuse New York, and the Center for Human Services Research at the University at Albany. As of February 2011, HFNY was operating in 37 sites throughout New York State.
Through community health and social service agencies, and hospitals, the HFNY program screens expectant parents and parents with an infant less than three months old for risk factors that are associated with child abuse, neglect, and poor developmental outcomes, including, but not limited to: single parenthood, teen pregnancy, poverty, poor education, unstable housing, substance abuse, and mental health problems. Parents who screen positive are referred to the HFNY program, which conducts an assessment interview to determine their eligibility using the Family Stress Checklist, a tool that measures the mother's risk of abusing or neglecting her children. Parents who score above a pre-determined cutoff on the Family Stress Checklist are offered the opportunity to receive home visitation services.
The researchers evaluated the HFNY program in a sample of 1,173 at-risk women who were pregnant or who had an infant less than three months old (DuMont et al., 2008). Families were selected for the evaluation based on the same criteria used to determine eligibility for HFNY. Women who were pregnant or who had an infant less than three months old in New York's Erie, Rensselaer, and Ulster counties were screened and then assessed for risk of child abuse and neglect. A total of 1,254 women met the eligibility criteria for the program and agreed to participate in the study. Among those women, 1,173 (94 percent) completed an intake interview and became study participants. The sample is comparable to the population of families enrolled in HFNY programs across the state on a range of measures such as demographics, employment, prior child abuse and neglect reports, substance use, and depression. Just over half of the participants were first-time mothers. Nine percent of the sample had substantiated child abuse or neglect reports prior to intake, and 20 percent had been physically abused in the previous year by a partner or spouse. About two out of every five participants also scored above the clinically relevant cutoff point on a depression scale.
A total of 579 women were randomly assigned to the treatment group (i.e., the HFNY group) and 594 were randomly assigned to the control group. Nearly two-thirds of the women were pregnant at the time they were assigned to the treatment or control group. The control group was given information and referrals to other appropriate services available in the community. Baseline statistics on the measures mentioned above showed no significant differences between the HFNY group and the control group, one indication that the random assignment was successful.
DuMont et al. (2008) evaluated the effects of HFNY on child abuse and neglect during the first two years of life. Of the 1,173 participants, 1,060 (90 percent of the sample) completed the Year 1 follow-up interview, and 992 (85 percent of the sample) completed the Year 2 follow-up interview. To maintain data quality, interviewers were independent of the HFNY program and were not informed of a participant's group assignment. Attrition analyses showed no significant differences between the HFNY group and the control group at the time of the Year 1 and Year 2 interviews. However, HFNY and control group members who remained in the study (regardless of whether they continued to participate in the HFNY program) were more likely than those who dropped out of the study to be first-time mothers, but less likely to have been randomly assigned at a gestational age of 30 weeks or less. In addition, at Year 1, rates of attrition from the study were slightly lower for non-Latina white women than for Latina women and non-Latina African-American women. However, by Year 2 these differences had disappeared. No other significant differences were found between those who continued in the study and those lost to attrition.
DuMont et al. (2008) conducted analysis on the Year 1 and Year 2 follow-up samples, using statistical adjustments to control for demographic and other risk-related factors. The researchers examined the outcomes of the entire sample and also two subgroups: "a high prevention opportunity" (HPO) subgroup (i.e., first-time mothers under age 19 who were randomly assigned at a gestational age of 30 weeks or less, about 15 percent of the sample) and a "psychologically vulnerable" subgroup: (i.e., women with a low sense of mastery and high levels of depressive symptoms at baseline, about 10 percent of the sample). Both subgroups included women in the HFNY group and the control group.
The study reported retention rates for the home visiting program consistent with those for other HFA programs. Specifically, at baseline, about 90 percent of the women who were assigned to the intervention group enrolled in the HFNY program, and only one-half and one-third of the HFNY participants remained in the program by Year 1 and Year 2, respectively. However, to the extent possible, those who dropped out of the HFNY program were retained in the intervention group of the study.
Lee and colleagues (2009) reported on HFNY's ability to impact low birth weight. The researchers obtained birth certificate data from the New York State Department of Health for mothers who were randomly assigned prenatally and provided informed consent. The match was performed using the names and dates of birth of the child and the mother, and was successful for 99 percent of the cases searched. Analyses were conducted on mothers who had a single birth and were randomized at a gestational age of 30 weeks or less in order to allow enough time prior to the target child's birth for participants to benefit from home visiting services.
Rodriguez et al. (2010) examined the effects of HFNY on the promotion of parenting competencies and the prevention of harsh parenting at the Year 3 follow-up, which included videotaped observational assessments of mother and child interactions. At two of the study sites, all of the mothers who met eligibility criteria were asked to participate in the Year 3 observational assessments. At the largest study site, a random sample of eligible mothers was asked to participate in the Year 3 observations. In order to be eligible for these observational assessments, the mothers had to have custody of their child, live within a two-hour driving distance of the interview location, and have completed the Year 2 interview. Of 643 selected mother-child pairs, 522 (81 percent) participated in the observations. This sample was comparable to the overall sample on most baseline characteristics, with the exception of a few significant differences between mothers who participated in the observational assessments and mothers who did not participate, in part due to the substantially reduced representation of participants from the largest (inner-city) site. Mothers who participated in the observations were less likely to be African-American, less likely to be Latina, more likely to be white, less likely to have been receiving welfare at baseline, and more likely to have healthy parenting attitudes at baseline. Mothers in the HFNY group were more likely to receive welfare at baseline, and more likely to report having a partner or spouse at baseline, when compared with the control mothers. These differences were between HFNY and control mothers were adjusted for using statistical controls.
The observational assessments were administered in participants’ homes by trained interviewers, who were independent of the HFNY program and blind to group assignment. Mothers and children were observed and videotaped while they interacted on three tasks: Puzzle Problem Solving, Delay of Gratification, and Cleanup. In the puzzle task, the mother is asked to work with the child on two puzzles, one of which is difficult for the child’s age, and the mother is asked to provide instruction when necessary. In the delay task, a snack is placed in front of the child and the mother is asked to fill out a questionnaire while also making sure the child does not eat or touch the snack. In the cleanup task, the mother and child play with toys for a time and then the mother is asked to have the child clean up the toys while she holds the bag. The data were also analyzed for the "high prevention opportunity" (HPO) subgroup, which included first-time mothers who were under the age of 19 and were randomized at a gestational age of 30 weeks or less.
DuMont et al. (2010) reported the findings from the Year 7 interview, including prevention of child maltreatment and risks for delinquency. The Year 7 follow-up included in-home interviews of mothers and the target children. Mothers were eligible to participate in this interview if both they and the target child were still alive, and if women in the control group had not received any HFNY services between the time of randomization and two weeks prior to the Year 7 interview. Target children also had to live within driving distance of an interviewer and had to be under their mother’s care and custody. Of the 1,173 original participants, 942 (80 percent) completed the Year 7 follow-up interview. Interviews were also conducted with 800 (71 percent) of the 1,128 children who were eligible for the Year 7 interview. Respondents and non-respondents at Year 7 were similar on the majority of baseline characteristics, with a few exceptions. Respondents were more likely than non-respondents to be African-American (48 percent vs. 36 percent), less likely to be Latina (16 percent vs. 28 percent), less likely to report moving in the prior year (54 percent vs. 63 percent), and presented with higher risk for child abuse or neglect initially as assessed by the Kempe Family Stress Checklist (5.77 vs. 5.39). The HFNY and control groups within the Year 7 mother and child samples were comparable on most attributes at baseline, but a few significant differences were found. Significantly more mothers interviewed had target children who were male in the HFNY group than in the control group (57 percent vs. 50 percent), mothers in the HFNY group had significantly lower annual earnings ($4,887 vs. $6,603) and were less likely to have a high school degree or equivalent (44 percent vs. 51 percent), and HFNY mothers were assessed with a higher level of risk for child abuse and neglect based on the Kempe Family Stress Checklist (5.86 vs. 5.68). Mothers of interviewed children in the HFNY group had significantly lower earnings ($4,733 vs. $6,190) and were less likely to have a high school degree or equivalent (44 percent vs. 51 percent). These differences between the HFNY and control group participants were adjusted for using statistical matching controls.
Key Evaluation Findings
The study has reported overall positive program findings in terms of childbirth outcomes (Lee et al., 2009) and parenting practices (DuMont et al., 2008; Rodriguez et al., 2010; DuMont et al., 2011). Compared with their counterparts in the control group, HFNY mothers were less likely to deliver low birth weight babies, less likely to engage in abusive, neglectful, or harsh parenting practices, and more likely to use positive parenting skills.
According to standard practice in the public health field, Lee and colleagues (2009) considered babies born weighing less than 2,500 grams to be low birth weight. Focusing on women who had a single birth and were randomized into the study at a gestational age of 30 weeks or less, Lee et al. (2009) reported significant findings based on birth certificate data provided by the New York State Department of Health:
- HFNY group mothers were only about half as likely as control group mothers to deliver low birth weight babies (5.1 percent vs. 9.8 percent). Rates were measured while controlling for other factors including race/ethnicity, program site, welfare receipt, age, and smoking during pregnancy.
- The program was particularly effective in reducing low birth weight among African-American and Latina mothers, groups that persistently experience high levels of poor birth outcomes. For example, African-American mothers who were assigned to the HFNY group were 70 percent less likely than African-American mothers in the control group to deliver low birth weight babies (3.1 percent vs. 10.2 percent).
- The earlier in their pregnancies that women were offered HFNY, the greater the impact of the program on low birth weight. Among women who entered the study at a gestational age of 16 weeks or less, the rate of low birth weight experienced by the HFNY group was one-quarter as high as the rate for the control group (3.6 percent vs. 14.1 percent)
- Compared with mothers in the control group, HFNY participants reported committing fewer acts of very severe physical abuse (0.01 vs 0.08), minor physical aggression (2.40 vs. 3.46), harsh parenting in the past week (1.21 vs. 1.81), and psychological aggression against their children (3.34 vs. 4.74) (DuMont et al., 2008).
- Among mothers in the psychologically vulnerable subgroup, the frequency of psychological aggression was significantly lower for the HFNY group (1.95) than for the control group (8.57) (DuMont et al., 2008).
- No significant differences were observed between the HFNY and control groups with respect to the rate or average number of substantiated CPS abuse and neglect reports. The evaluators note that the discrepancy in the findings for official indicators of abuse and neglect and self-reported measures may be due to greater surveillance of HFNY parents by home visitors and the providers to which they refer families. HFNY parents who admitted to having committed acts of serious abuse and neglect were nearly twice as likely to have a CPS report than were control parents who self-reported serious abuse or neglect, suggesting that actual incidents of abuse and neglect committed by HFNY parents were more likely to be detected and reported to CPS than were those committed by control parents (Mitchell-Herzfeld et al., 2005).
- No program effects on the prevalence or frequency of substantiated CPS reports were detected at Year 2.
- The average number of reported acts of severe physical abuse in the past year was lower among mothers in the HFNY program than mothers not participating in the program (0.01 vs. 0.04).
- Among mothers who were "psychologically vulnerable" (mothers who had a low sense of mastery and high levels of depressive symptoms), mothers in the HFNY program were less likely to report acts of serious abuse or neglect (5 percent vs. 19 percent). Among psychologically vulnerable mothers, the average number of self-reported incidents of serious abuse and neglect in Year 2 was significantly lower for the HFNY group (0.02) than for the control group (0.62).
- Among mothers in the "high prevention opportunity" (HPO) subgroup (first-time mothers under age 19 who were randomized at a gestational age of 30 weeks or less), mothers in the HFNY program were less likely to report acts of minor physical aggression in the past year (51 percent vs. 70 percent), and harsh parenting behaviors in the past week (41 percent vs. 62 percent).
- Significantly more HFNY mothers used positive parenting strategies (e.g., maternal responsivity and cognitive engagement) during all three tasks (Puzzle: 95.5 percent vs. 92.8 percent, Delay: 17.2 percent vs. 10.6 percent, Cleanup: 85.3 percent vs. 78.3 percent).
- Among the overall sample, there were no significant differences between mothers who participated in the program and mothers who did not participate in the program on harsh parenting or role-reversed parenting.
- Mothers in the "high prevention opportunity" (HPO) subgroup who participated in the program were significantly less likely to use harsh parenting during the Puzzle and Delay tasks (Puzzle: 5.3 percent vs. 21.5 percent, Delay: 5.3 percent vs. 23.8 percent) than HPO subgroup mothers who did not participate in the program.
At the Year 7 interview, DuMont et al (2010) found the following:
- For the sample as a whole, there was no difference between intervention and control parents in the rate or cumulative number of CPS abuse or neglect reports.
- Mothers in the HFNY group reported using serious physical abuse less frequently in the past year than mothers in the control group (0.03 vs. 0.15).
- Mothers in the HFNY group reported using non-violent discipline strategies more frequently in the past year than mothers in the control group (an average of 49 times vs. 45 times) and reported higher rates of using non-violent discipline (100 percent vs. 98.6 percent).
- Children in the HFNY group were less likely to report that their mothers engaged in minor physical aggression against them than children in the control group (70.1 percent vs. 77.2 percent).
- Significantly more children who took part in the HFNY program participated in gifted programs within their schools than children in the control group (5.38 percent vs. 1.99 percent).
- Intervention children were less likely to report skipping school, but, according to maternal reports, there was no difference in skipping school.
- There were no significant differences in problem behavior, socio-emotional difficulties, or self-regulation.
- Among children in the HPO subgroup, children in the HFNY group were significantly less likely to score below average on the PPVT-IV receptive vocabulary assessment (59.40 percent vs. 77.61 percent).
- Compared with mothers in the RRO subgroup who did not participate in the program, mothers in the RRO subgroup who participated in the program had lower rates of initiation of child welfare cases for preventative, protective, or placement services (38.02 percent vs. 60.02 percent).
- HFNY mothers in the RRO subgroup had a smaller number of total confirmed CPS reports for mothers than the control subjects (0.8 vs. 1.6).
- No significant differences were found among mothers in the HPO subgroup on confirmed CPS reports of abuse or neglect, reports of physical abuse, or initiation of preventative, protective, and placement services.
Public health and social welfare services
The project is funded through the state budget, and the allocation for the program was $23.3 million in 2010. The cost per family ranges from $3,500 to $4,500 per year, with slightly higher costs in New York City.
HFNY is a variant of the HFA home visiting model, and this variant is shown to have significant and positive impacts. HFNY participants may receive home visiting services until the child reaches the age of five or is enrolled in Head Start or kindergarten. Families are served at different service levels that correspond to different frequencies of home visits based on families' needs. Home visits are scheduled one or more times per week during pregnancy (Level 1), and families usually remain on Level 1 until the child is at least six months old. As families progress through the service levels, home visits occur on a diminishing schedule, from biweekly (Level 2), to monthly (Level 3), and then quarterly (Level 4). The content of home visits is individualized and culturally appropriate to address the unique needs of each family.
HFNY is managed by the New York State Office of Children and Family Services, which contracts with public and community-based agencies to provide home visitation services. Funded programs are required to follow HFA and HFNY standards and participate in the HFA credentialing process.
All new home visitors attend a one-week core training program provided by approved in-state HFA trainers. They receive training on parent-child interaction, child development, strength-based service delivery, and on such a range of topics such as domestic violence, substance abuse issues, abuse and neglect, and well-baby care. HFNY supervisors receive an additional four days of training on their role in promoting quality services. New home visitors are mentored by experienced home visitors. Supervisors meet with each home visitor for at least 1.5 hours every week and observe one home visit per quarter.
The program selects paraprofessionals from the community being served to be home visitors. Personal attributes such as warmth, the ability to establish trusting relationships, the ability to work effectively with children and families, and nonjudgmental attitudes are the primary selection criteria. Although post-secondary education is not a requirement, many home visitors (43 percent) have attended college, and about one-third (34 percent) are college graduates.
Issues to Consider
The HFNY program received a "proven" rating. The study utilized random assignment with a fairly large sample size of 1,173 women. To maintain data quality, interviewers were independent of the HFNY program and were not informed of a participant's group assignment. Parents who were randomly assigned to the HFNY program were kept in the analysis, regardless of whether or not they actually received any home visiting services. DuMont et al. (2008) reported a 10 percent attrition rate at Year 1 and a 15 percent attrition rate at Year 2; 2 percent of the study sample completed the Year 2 but not the Year 1 interview. DuMont et al. (2010) reported a 20 percent attrition rate at Year 7. At Years 1, 2, and 7, attrition analyses indicated no significant differences between mothers in the control and HFNY groups with respect to attrition. However, the analyses at all three years indicated significant differences in baseline measures between parents lost to attrition and those retained in the sample. At Year 1 differences were found on race and parenting attitudes, at Year 2 differences were found on randomization before 30 weeks of gestational age and being a first time mother, and at Year 7 differences were found on race, report of moving in the past year, and initial risk of child abuse or neglect. At Year 3, a reduced sample of mothers were observed interacting with their children on a series of semi-structured tasks. All eligible mothers who were initially interviewed at two study sites were asked to participate in the observational assessments, and at one study site a random subsample of eligible mothers were selected. There were significant differences between mothers who participated in the Year 3 interview and mothers who did not participate.
The study found significant and positive program effects on a range of outcomes, including child abuse and child neglect and childbirth outcomes. Although, some of the program effects found in Year 1 did not persist into Year 2, the reduction in the average number of maternal-reported acts of severe physical abuse found at Year 1 and Year 2 was sustained at Year 7.
Dozens of evaluations of HFA have been conducted, but most lacked a comparison group or used quasi-experimental designs, and most failed to find significant positive effects. Possible reasons for the evaluations not finding significant effects could be the relatively small sample sizes and a low base rate for child abuse and neglect reports. The studies summarized here are the only studies of an HFA model that have met the Promising Practices Network (PPN) evidence criteria, including such factors as study design, effect size, and statistical significance.
HFNY is very similar to many other HFA programs in that it satisfies all the critical elements required by the HFA model, including participation eligibility, service content, home visitor selection and training, and service monitoring and supervision. HFNY differs from other programs only in some program details that are allowed under the HFA structure.
Previous studies have shown larger program effects from the Nurse Family Partnership (NFP) home visiting model (see PPN description at http://www.promisingpractices.net/program.asp?programid=16) than the HFA model. One hypothesis forwarded to explain the differences focuses on the training of the provider of the home visitation services —the NFP model uses nurse home visitors, whereas the HFA model employs paraprofessional home visitors. However, DuMont et al. (2008) found more pronounced program effects among the first-time mothers under age 19 than for the entire sample, and this subgroup of first-time mothers resembled the type of participants typically served by NFP programs. Therefore, the HFNY evaluation proposed that the characteristics of recipients might be a key factor in explaining the differences in the program effects between the NFP and HFA model. In fact, consistent with these findings from the HFNY subgroup analysis, the NFP data have shown that higher-risk mothers also benefit more from NFP services (see the discussion in Karoly et al., 2001). Further studies that examine this hypothesis would be valuable.
A cost-benefit analysis of HFNY over the first seven years of life found that the program did not pay for itself through generation of government savings in public assistance and food stamp payments, Medicaid-funded births, CPS investigations, preventative services, and foster care placements. For the sample as a whole, for every dollar spent on the program, $0.15 were returned in savings to the government due to decreased use of government supported programs. Among women in the recurrence reduction opportunity (RRO) subgroup, the program generated a savings of $3.16 for every dollar spent on the program, and among women in the high prevention opportunity (HPO) subgroup, the recovery was $0.25 in savings for every dollar spent on the program.
One of the members of the study team, Susan Mitchell-Herzfeld, is on the PPN Board of Advisors.
Erie, Rensselaer, and Ulster Counties in New York State
Healthy Families New York
New York State Office of Children and Family Services
52 Washington Street, 334N
Rensselaer, NY 12144
Fax: (518) 402-6824
Healthy Families New York website:
New York State Office of Children and Family Services website:
Center for Human Services Research, University at Albany website:
DuMont, K., K. Kirkland, S. Mitchell-Herzfeld, S. Ehrhard-Dietzel, M. Rodriguez, E. Lee, C. Layne, and R. Greene,
Final Report: A Randomized Trial of Healthy Families New York (HFNY): Does Home Visiting Prevent Child Maltreatment?
2010. As of February 2011:
DuMont, K., S. Mitchell-Herzfeld, R. Greene, E. Lee, A. Lowenfels, M. Rodriguez, and V. Dorabawila, "Healthy Families New York (HFNY) Randomized Trial: Effects on Early Child Abuse and Neglect," Child Abuse and Neglect, Vol. 32, 2008, pp. 295-315.
Lee, E., S. Mitchell-Herzfeld, A. Lowenfels, R. Greene, V. Dorabawila, and K. DuMont, "Reducing Low Birth Weight Through Home Visitation: A Randomized Controlled Trial," American Journal of Preventive Medicine, Vol. 36, No. 2, 2009.
Mitchell-Herzfeld, S., C. Izzo, R. Greene, E. Lee, and A. Lowenfels, Evolution of Health Families New York (HFNY): First Year Program Impacts, 2005. As of February 2011: http://www.ocfs.state.ny.us/main/reports/#hfny
Rodriguez, M. L., K. Dumont, S. Mitchell-Herzfeld, N. J. Walden, and R. Greene, "Effects of Healthy Families New York on the Promotion of Maternal Parenting Competencies and the Prevention of Harsh Parenting," Child Abuse and Neglect, Vol. 34, 2010, pp. 711-723.