Journal Issue: Preventing Child Maltreatment Volume 19 Number 2 Fall 2009
Outcomes of Home-Visiting Programs
Although the focus of this volume of The Future of Children is preventing child abuse and neglect, we will review the outcomes of several home-visiting programs in multiple domains. In addition to child abuse and neglect, we will also discuss outcomes related to child health and safety, parenting, maternal mental health, and children’s cognitive development. Unfortunately, few studies have documented effects on reducing or preventing child abuse and neglect. However, given the association between certain aspects of parenting and child outcomes (as we discussed earlier), measures of parenting and maternal and family functioning may shed important insights on child well-being.
Child Abuse and Neglect
As noted, assessing the prevalence of child abuse and neglect involves a number of difficulties, such as varying definitions, low reporting rates, and the difficulties of substantiating cases. As a result, research is generally weak in this area. Some programs, however, such as the NFP, HSP, HFA, and Early Start, have specifically examined abuse and neglect as outcomes of the program, and some have shown positive effects in this domain. Perhaps the most widely cited finding from a home-visiting program was based on the Elmira evaluation of the NFP, which documented a 48 percent decline in rates of child abuse and neglect at the time of the fifteen-year follow-up among low-income families who had received the intervention.48 Other studies that have attempted to examine Child Protective Services reports of abuse and neglect as an outcome measure have also found low prevalence rates in both groups, resulting in low power to detect statistically significant differences. Neither HSP nor any of the randomized HFA evaluations have identified significant reductions in substantiated cases of child abuse or neglect as a result of their programs, though the Alaska evaluation did note a significant reduction in CPS referrals (from 73 to 42 per thousand over a two-year period).49 Typically, rates of child abuse and neglect were low across both groups. For example, Healthy Families New York identified that 6 percent of the controls and 8 percent of the treatment group had substantiated reports of abuse or neglect at one year. At two years, the rates were around 5 percent for both groups. Neither the one- or two-year data yielded any significant differences between families in the treatment and control groups.50 Early Start also examined CPS referrals and substantiated cases and found no differences for either measure between treatment and control families—21 percent of control families had contact with CPS agencies, compared with 20 percent of program families.51
Another strategy for gauging the rates of child abuse and neglect—asking parents directly about their own behaviors toward their children—yields more promising results. The evaluation of HFNY found many significant links between program involvement and reductions of abusive or neglectful behaviors, though few were observed at both one and two years. At one year, but not at two years, mothers in the program group engaged less frequently in acts of psychological aggression.52 In contrast, neglectful behaviors53 did not differ at one year, but did at two years. Effects were more consistent on physical abuse, however, with mothers in the treatment group reporting fewer instances of very serious physical abuse at one year and fewer instances of serious abuse at two years.54 In Alaska, the HFA program was associated with less psychological aggression, but it had no effects for neglect or severe abusive behaviors.55 Similarly, in the San Diego evaluation of HFA, home-visited mothers reported less use of psychological aggression at twenty-four and thirty-six months.56 Early Start also reported small effects in terms of lowering rates of severe physical abuse.57
In contrast, Hawaii Healthy Start showed no overall effects in terms of parent-reported abusive or neglectful behaviors, even though the program was initially designed to prevent child abuse and neglect. Overall, the treatment and control groups differed little with respect to child abuse and neglect. Only two differences emerged: HSP mothers were less likely to use corporal or verbal punishment or engage in neglectful behaviors. In both cases, the effects were isolated within a single site (not the same site for both effects). Overall, the authors concluded that the program did little to prevent child abuse.58 They also noted that the home visitors rarely expressed concerns about child maltreatment, even among families for whom other measures suggested significant problems.
Relatively few home-visiting studies have collected adequate measures of child abuse and neglect. As noted, those that attempt to assess effects in this domain often yield inconclusive results. The problem, however, may simply be that the low overall prevalence of documented cases of abuse and neglect makes it almost impossible for most clinical trials to detect significant changes in this domain. Furthermore, mothers who are in programs may be more likely to be detected and receive services for suspected abuse or neglect. As a result, additional child and parent measures are necessary to understand fully the effect of home-visiting programs on family and child well-being.
Harsh Parenting Behaviors
Harsh parenting behaviors are those on the milder end of the continuum of abusive behaviors. In contrast to indices of abuse and neglect, harsh parenting is evidenced by things like spanking, slapping, or pinching the child.59 The Healthy Families New York evaluation examined a number of harsh parenting behaviors in addition to their measures of abuse and neglect. They found evidence that families in the intervention group exhibited fewer harsh parenting behaviors than families in the control group and that this effect was particularly strong among first-time mothers who had enrolled in the program during pregnancy (62 percent of controls vs. 41 percent of the treatment group). Among the prevention subgroup (first-time mothers recruited prenatally), minor physical aggression was reported in 70 percent of control families and 51 percent of program families.60 In Healthy Families Alaska, fewer incidents of mild physical abuse were reported among families in the treatment group.61
The Nurse-Family Partnership has also shown positive effects in reducing harsh parenting behaviors among adolescent mothers. In the Elmira demonstration, intervention mothers were less likely to punish or physically restrain their children than mothers in the control group.62 Among home-visited families who participated in Early Start, less punitive parenting was observed, though the effect was modest.63 Several other programs have identified reductions in the frequency with which mothers spanked their children at thirty-six months, including Healthy Families San Diego,64 Early Head Start,65 and IHDP.66 No effects on harsh parenting were found in the CCDP.67
Child Health and Safety
Aspects of children’s health and safety such as the number of injuries and hospital admissions, as well as immunizations and doctor and dental visits, can provide important insight into a child’s quality of care. Accordingly, a number of home-visiting evaluations have measured outcomes in this domain.
The NFP examined both injuries and hospital admission in the Elmira and Memphis evaluations. In Elmira, children of low-income, unmarried mothers in the treatment group had fewer emergency room visits than controls.68 Similarly, in Memphis, fewer accidents and injuries required treatment. In the Memphis site, nurse-visited families also had lower child mortality. One child in the treatment group died, compared with ten in the control group.69
Several studies have examined the effects of home visiting on children’s completion of immunizations, though few have identified program benefits in this area. Of those that examined immunizations (NFP-Memphis, HFA, HSP, EHS, Queensland, and Early Start), only EHS identified a significant program effect on immunizations, though the size of the effect was quite small and applied to the comparison of the entire treatment group to controls, not specifically to those families who had received home visits.70 The one-year follow-up of the Queensland program also suggested a trend in favor of the intervention group’s having higher levels of vaccinations than the control group.71
The Early Start program in New Zealand was one of the few evaluations to identify effects on the frequency of doctor and dental visits. Families in the program group had more general practitioner visits over thirty-six months, a higher proportion were up to date with well-child checks, and they were more likely to have had dentist visits.72 The Queensland program and Hawaii Healthy Start both examined the number of well-child visits and found no differences across groups. Furthermore, neither HSP nor any of the three HFA evaluations identified effects in terms of linking program families to a medical home.73
Quality of the Home Environment
More programs have observed positive effects in parenting domains than in child outcomes. With regard to the quality of the home environment,74 several programs have identified positive effects. For example, the Queensland study documented higher-quality home environments for families in the intervention.75 Likewise, positive effects were observed on measures of the home environment in Alaska.76 Among multi-component programs, both Early Head Start77 and the Infant Health and Development Program78 reported higher-quality home environments in the intervention groups, though effect sizes tended to be small. In contrast, the CCDP did not significantly affect the home environment or any measured aspects of parenting.79
A conflicting picture emerged from the results of the Nurse-Family Partnership across the three evaluation sites. In Denver, mothers who received home visits had more sensitive mother-infant interactions and higher HOME scores than mothers who did not.80 Home visiting, however, had no significant effects on different aspects of the home environment in Elmira or Memphis.81 One possible explanation for this difference is that the majority of mothers at the Elmira and Memphis sites were adolescents, whereas the Denver mothers were more diverse in age, suggesting stronger effects for older mothers than for younger mothers with respect to the quality of the home environment.
Increased Parenting Responsivity and Sensitivity
As several studies have documented, home-visiting programs are often associated with parental gains in responsivity and sensitivity in their interactions with their children. In the Infant Health and Development program, mothers in the intervention group engaged in higher-quality interactions with their infants, though the effects were small.82 In New Zealand, Early Start documented higher positive parenting attitudes, a greater prevalence of nonpunitive attitudes, and more favorable overall parenting scores for families in the treatment group.83 In Queensland, mothers in the intervention group were rated as significantly higher in emotional and verbal responsivity.84
Evidence also shows that home-visiting programs can improve maternal parenting sensitivity. The Netherlands program, for example, achieved its primary goal—improving maternal sensitivity. At the end of the study, mothers who had received home visits were more sensitive in their interactions with their infants and more skilled in structuring activities with the child.85 Other home-visiting programs with broader aims have also identified program effects on maternal sensitivity. Home-visited mothers in the Denver site of the NFP were rated as more sensitive during interactions with their children. The effect was small, but was identified in the whole program group, instead of only in a smaller subgroup.86 In Memphis, more positive interactions were observed in the subgroup of women who possessed low psychological resources.87 Likewise, home-visited mothers in Early Head Start were rated as more supportive during play with their children than controls, though the effect was small.88 Maternal sensitivity was also examined in Hawaii Healthy Start, the Healthy Families evaluations in San Diego and Alaska, and the Comprehensive Child Development Program, though none identified significant effects.
Maternal Depression and Parenting Stress
Some programs have examined depressive symptoms and parenting stress as outcomes of the intervention. One evaluation conducted in Queensland, Australia, reported moderate reductions in depressive symptoms for mothers in the intervention group at the six-week follow-up.89 A subsequent follow-up, however, suggested that these benefits were not long lasting, as the depression effects had diminished by one year.90 Similarly, Healthy Families San Diego identified reductions in depression symptoms among program mothers during the first two years, but these effects, too, had diminished by year three.91 In Healthy Families New York, mothers at one site (that was supervised by a clinical psychologist) had lower rates of depression at one year (23 percent treatment vs. 38 percent controls).92 The Infant Health and Development program also demonstrated decreases in depressive symptoms after one year of home visiting, as well as at the conclusion of the program at three years.93 Among Early Head Start families, maternal depressive symptoms remained stable for the program group during the study and immediately after it ended, but decreased just before their children entered kindergarten.94 No program effects were found for maternal depression in the Nurse-Family Partnership, Hawaii Healthy Start, Healthy Families Alaska, or Early Start programs.
Some effects on parenting stress have also been identified. Most notably, home-visited families participating in Early Head Start reported experiencing significantly less stress in their parenting roles than did control families.95 The same pattern occurred in Queensland: mothers who received home-visiting services reported less stress in the parenting role than did mothers in the control group.96 Healthy Families programs in Alaska, San Diego, and Hawaii also examined parenting stress in their evaluations. In Alaska, 22 percent of families who received HFA services reported very high levels of parenting stress (above 90th percentile), as compared with 30 percent of mothers in the control group. In San Diego, a small effect was noted in favor of treatment families’ having lower stress, but the relationship was only marginally significant. Hawaii Healthy Start did not yield any effects on parenting stress.97
Another interesting approach is to focus on mothers who are clinically depressed as targets for the intervention. In the Netherlands program, all mothers were receiving outpatient psychotherapy for their depression. Accordingly, mothers in both groups showed reductions in depressive symptoms over the course of the study. However, there were no additional benefits for mothers in the treatment group.98
Overall, this pattern of results suggests that home-visiting programs may not be designed to handle problems associated with high levels of stress or mental illness, which may be best treated in other settings. Although depressed mothers may gain parenting skills as a result of home intervention programs, they are unlikely to feel less parenting stress or fewer depressive symptoms per se. This important finding shows that the effectiveness of home-visiting programs is limited and that those that have well-defined goals in certain domains are most likely to evidence effects. At the same time, it is worth noting that some programs did identify small effects on stress and depressive symptoms and that others have specifically targeted reducing maternal depressive symptoms and have obtained stronger results.99
Children’s Cognitive Development
Effects on children’s cognitive development have been more difficult to identify in home-visiting programs, largely because the programs rarely provide services directly to children. Because effects on parenting are modest, it follows that effects on children would be even smaller. Even so, there is some evidence that changes in children’s outcomes are mediated by changes in parenting attitudes and behaviors.100
In Hawaii Healthy Start and the CCDP, no cognitive benefits were observed for children. However, in Healthy Families Alaska, program children had higher Bayley scores at age two than controls, with 58 percent of intervention children and 48 percent of controls scoring in the normal range.101 In the Nurse-Family Partnership evaluations, some effects were observed within each of the three evaluations, but most effects were concentrated within specific subgroups of families. In Denver, low-resource families who received home visiting showed modest benefits in children’s language and cognitive development.102 In Elmira, only the intervention children whose mothers smoked cigarettes before the experiment experienced cognitive benefits.103 In Memphis, children of mothers with low psychological resources104 in the intervention group had higher grades and achievement test scores at age nine than their counterparts in the control group.105 Early Head Start also identified small, positive effects on children’s cognitive abilities, though the change was for the program as a whole and not specific to home-visited families.106 Similarly, IHDP identified large cognitive effects at twenty-four and thirty-six months, but not at twelve months, so the effects cannot be attributed solely to home-visiting services.107
Summary of Outcomes
Table 2 summarizes the results of the home-visiting programs just described. In general, a review of the literature reveals a mixed picture regarding the efficacy of home-visiting programs. In each domain, some studies have documented effects whereas others have not. Furthermore, many effects are isolated within specific subgroups of families or within individual sites, so that findings cannot be generalized to the entire population served. In an attempt to reconcile these disparate and often contradictory findings, several researchers have undertaken meta-analyses to estimate effects across a number of programs. Often, these meta-analytic reviews include both experimental evaluations (randomized controlled trials) and quasi-experimental evaluations, whereas we feel that conclusions should be based primarily —if not entirely—on experimental evaluations. Even so, the results of meta-analyses can be instructive.
Monica Sweet and Mark Appelbaum published a meta-analysis that included sixty home-visiting programs (including both quasi and true experiments). They found evidence that home visiting is associated with benefits in parenting attitudes and behavior, as well as in children’s cognitive development. However, for both child abuse and parent stress, the average effect sizes were not different from zero, suggesting a lack of evidence for effects in these areas.108 Earlier meta-analytic reviews have also noted the lack of sizable effects in preventing child maltreatment—again citing the different intensity of surveillance of families in the treatment versus control groups as an explanation (though the authors did report that home visiting was associated with an approximately 25 percent reduction in the rate of childhood injuries).109 Another review focusing on the quality of the home environment also found evidence for a significant overall effect of home-visiting programs.110 More recently, Harriet MacMillan and colleagues published a review of interventions to prevent child maltreatment, and identified the Nurse-Family Partnership and Early Start programs as the most effective with regard to preventing maltreatment and childhood injuries. The authors note that many other programs lack strong evidence of such effects.111
Taken together, these findings suggest that home-visiting programs offer little evidence that they directly prevent child abuse and neglect. The evidence, however, is stronger with respect to parenting and the quality of the home environment. Study findings show that home visits can impart positive benefits to families by way of influencing maternal parenting practices, the quality of the child’s home environment, and children’s development. And because other studies have linked parenting quality with child maltreatment, improved parenting skills would likely be associated with improved child well-being and corresponding decreases in maltreatment, even if these effects remain difficult to document.
Another tool for considering the effectiveness of intervention programs is cost-benefit analysis. Although few such analyses have been conducted with home-visiting programs, some interesting findings have nevertheless emerged. The Elmira site of the Nurse-Family Partnership has been evaluated on two separate occasions, originally by Lynn Karoly and colleagues at RAND and again by Steve Aos at the Washington State Institute for Public Policy.112 In both analyses, benefits tended to outweigh costs. Savings were primarily in four areas: increased tax revenues associated with maternal employment, lower use of public welfare assistance, reduced spending for health and other services, and decreased criminal justice system involvement. For the higher-risk group in Elmira, each dollar invested yielded $5.70 in savings. For the lower-risk group, the saving was $1.26 per dollar invested.113 For the full sample, Aos calculated an overall benefit-cost ratio of $2.88. The Aos evaluation also assessed the costs and benefits as reported in a meta-analysis of home-visiting programs and found an average of $2.24 saved for each dollar invested in home-visiting programs. A cost-benefit analysis of Healthy Families America, however, showed a net loss of 4.8 cents for each dollar invested in the program, and Early Head Start showed a net loss of 7.7 cents per dollar invested. Cost benefits would, of course, increase if longer-term follow-ups continued to show benefits of these programs.