The Effectiveness of Early Childhood Home … Effectiveness of Early Childhood Home Visitation in...

TVAOAA

O

IptHorcnatijhoTtpvigipcpt

I

Ea

FCaAloM

PPR

A©

he Effectiveness of Early Childhood Homeisitation in Preventing ViolenceSystematic Review

leg Bilukha, MD, PhD, Robert A. Hahn, PhD, MPH, Alex Crosby, MD, MPH, Mindy T. Fullilove, MD,kiva Liberman, PhD, Eve Moscicki, ScD, MPH, Susan Snyder, PhD, Farris Tuma, ScD, Phaedra Corso, PhD,
manda Schofield, MPH, Peter A. Briss, MD, MPH, Task Force on Community Preventive Services
vvwat

pdiccabcwefdp

gsfiad(accppe(ciafmwvb

verview

n early childhood home visitation programs, par-ents and children are visited at home during thechild’s first 2 years of life by trained personnel who

rovide some combination of information, support, orraining about child health, development, and care.ome visitation has been used to meet a wide range of

bjectives, including improvement of the home envi-onment, family development, and the prevention ofhild behavior problems. The Task Force on Commu-ity Preventive Services (the Task Force) has conductedsystematic review of scientific evidence of the effec-

iveness of early childhood home visitation for prevent-ng violence, with a focus on violence by and againstuveniles. The Task Force recommends early childhoodome visitation for preventing child abuse and neglect,n the basis of strong evidence of effectiveness. Theask Force found insufficient evidence to determine

he effectiveness of early childhood home visitation inreventing violence by visited children, violence byisited parents (other than child abuse and neglect), orntimate partner violence in visited families. This reportives additional information about the findings, includ-ng diverse outcome measures and results in studyopulation subsamples, describes how the reviews wereonducted, provides information that can help in ap-lying the intervention locally, and recommends addi-ional research.

ntroduction

arly childhood home visitation has been used toddress a wide range of public health goals for both

rom the Epidemiology Program Office (Bilukha, Hahn, Snyder,orso, Schofield, Briss) and National Center for Injury Preventionnd Control (Crosby), Centers for Disease Control and Prevention,tlanta, Georgia; Department of Psychiatry and Public Health, Co-

umbia University (Fullilove), New York, New York; National Institutef Justice (Liberman), Washington, DC; and National Institute ofental Health (Moscicki, Tuma), Bethesda, MarylandAddress correspondence and reprint requests to: Robert A. Hahn,

hD, Community Guide Branch, Centers for Disease Control and

srevention, 1600 Clifton Road, MS E-90, Atlanta, GA 30333. E-mail:[email protected].

m J Prev Med 2005;28(2S1)2005 American Journal of Preventive Medicine • Published by

isited children and their parents, including not onlyiolence reduction but also other health outcomes, asell as health-related outcomes such as educationalchievement, problem-solving skills, and greater accesso resources.1,2

In our review, “home visitation” is defined as arogram that includes visitation of parent(s) and chil-(ren) in their home by trained personnel who convey

nformation about child health, development, andare; offer support; provide training; or deliver anyombination of these services. Visits must occur duringt least part of the child’s first 2 years of life, but canegin during pregnancy and can continue after thehild’s second birthday. We allowed for programs inhich participation in home visitation programs wasither voluntary or mandated (e.g., by a court), butound no program in which participation was man-ated. Visitors can be nurses, social workers, otherrofessionals, paraprofessionals, or community peers.In the United States, home visitation programs have

enerally been offered to specific population groups,uch as low income; minority; young; less educated;rst-time mothers; substance abusers; children at risk ofbuse or neglect; and low birth weight, premature,isabled, or developmentally compromised infants.Home visitation programs are common in Europe andre most often universal [i.e., made available to allhildbearing families, regardless of estimated risk ofhild-related health or social problems]).3 Visitationrograms are often “two generational,”4 addressingroblems and introducing interventions of mutual ben-fit to parents and children. Programs may includebut are not limited to) one or more of the followingomponents: training of parent(s) on prenatal andnfant care; training on parenting to prevent childbuse and neglect; developmental interaction with in-ants and toddlers; family planning assistance; develop-

ent of problem-solving and life skills; educational andork opportunities; and linkage with community ser-ices. Home visitation programs may be accompaniedy the provision of day care; parent group meetings for
upport, instruction, or both; advocacy; transportation;
110749-3797/05/$–see front matterElsevier Inc. doi:10.1016/j.amepre.2004.10.004

aa“

boiowcmaafaicem

stacauaconreslctsvgmtSd

nltsoqEessf

1

2

34

btotcachairBlIeSrtsassSn

wthSbpiahv

t(pnmmpc

1

nd other services. When such services are provided inddition to home visitation, we refer to the program asmulticomponent.”

Several theoretical orientations indicate the potentialeneficial effects of home visitation on violence andther outcomes.5,6 Human ecology theory7 clarifies the

mportance of the social environment—including notnly the influence of parents, but also of social net-orks, neighborhoods, communities, and cultures—inhild development. Evidence shows that an environ-ent of community disorganization and poverty can besource of crime and violence.8 Home visitation is seens strengthening the capacities of parents in success-ully relating to their social environment and gainingccess to social resources. Because the effects of parent-ng are critical in the development and prevention ofhild violence,8 home visitors also teach effective par-nting and work to strengthen the support of familyembers and friends.Enhancing parents’ sense of self-efficacy also

trengthens their capacities as parents. The underlyingheory of self-efficacy is that people are more likely toct when they believe both that they are capable ofarrying out a given action and that this action willccomplish a desired goal.9 Home visitors may contrib-te here by encouraging and facilitating successful,chievable modifications in parents’ lives, possibly in-luding steps toward career development. Increasedccupational independence may provide not onlyeeded resources, but a sense of accomplishment andelief from stresses that distract from child care. Self-fficacy may also improve family planning and childpacing, thereby reducing maltreatment, which is moreikely with greater numbers of children and childrenlose to one another in age.10 Finally, attachmentheory11,12 stresses the importance of a close relation-hip with parents for healthy child development; homeisitors can play a role in strengthening attachment byiving guidance on effective parenting. Home visitorsay work to modify harmful patterns of relationship

hat were learned in the parents’ own upbringing.13

trong parental involvement can protect against theevelopment of child violence.8

The purpose of this review is to assess the effective-ess of home visitation programs in preventing vio-

ence. Therefore, we reviewed studies of home visita-ion only if they assessed violent outcomes. We reviewedtudies whether or not violence was the primary targetr outcome of the visitation, as long as the studyualified by specified inclusion criteria (see “Search forvidence” section) and assessed violent outcomes. Theffects on other outcomes were not systematically as-essed, but are selectively reported if addressed in thetudies reviewed. We reviewed studies examining any of
our violent outcomes: f
2 American Journal of Preventive Medicine, Volume 28, Num

. Violence by the visited child, against self or others,including violence in school, delinquency, crime, orother observed or reported violent behavior

. Violence by the visited parent, other than childmaltreatment

. Intimate partner violence

. Violence against the child, specifically maltreatment(which includes all forms of child abuse andneglect)

Violence in which juveniles are offenders, victims, oroth is a substantial problem in the United States. Overhe last 25 years, juveniles have been involved asffenders in at least 25% of serious violent victimiza-ions.10 Since at least 1976, the highest rates of homi-ide in the United States have occurred among peopleged 18 to 24 years.10 In 1994, 33% of juvenile homi-ide victims were killed by a juvenile offender. Rates ofomicide victimization among youth aged �15 yearsre five times higher in the United States than they aren the combination of other industrialized nations andegions for which data are available (Australia, Austria,elgium, Canada, Denmark, England and Wales, Fin-

and, France, Germany, Hong Kong, Ireland, Israel,taly, Japan, Kuwait, Netherlands, New Zealand, North-rn Ireland, Norway, Scotland, Singapore, Sweden,pain, Switzerland, and Taiwan). Rates of firearm-elated homicide are approximately 16 times higher inhe United States than in those same nations.14 Rates ofuicide also rise substantially during adolescence, reachplateau among people aged 35 to 44 years, and rise

ubstantially again only after age 65 years.15 The rate ofuicide among children aged �15 years in the Unitedtates is twice that of the combination of industrializedations noted above.14

Although intimate partner violence victimizes men asell as women in the United States, women are three

imes more likely to be victims than are men.16 Duringer lifetime, one out of four women in the Unitedtates will be the victim of partner violence: 7.7% wille victims of rape and 22.1% will be victims of otherhysical assaults.16 Violent victimization of women,

ncluding threats of rape and sexual assault, is greatestmong women aged 16 to 19 years. Such violence canave severe physical and mental consequences forictims.17

In 1999, 4.1% of children (aged �18) were reportedo be victims of maltreatment. Many of those reports33.8%) are investigated and not confirmed by childrotective services. Further complicating this picture,ational survey data indicate that additional cases ofaltreatment are not reported.10,18,19 Child maltreat-ent can include physical, sexual, or emotional abuse;

hysical, emotional, or educational neglect; or anyombination of these. Not only is child maltreatment a
orm of violence in and of itself, but it is associated with
ber 2S1

asmrbaebpaSp

T

TomFSUcCpmttFpaGtCt

H

TitftteMthotivtnsT

I

INhSt

Th

I

U

V

a

b

c

d

wC1m

dverse consequences among maltreated children,uch as early pregnancy, drug abuse, school failure,ental illness, and suicidal behavior.20,21 Although the

elationship is not well understood, children who haveeen physically abused are more likely to perpetrateggressive behavior and violence later in their lives,ven after accounting for other risk factors for violentehavior.22 Abuse and neglect are both associated withoverty and single-parent households; for reasons suchs these, many home visitation programs in the Unitedtates are directed to poorer, minority, and single-arent families.

he Guide to Community Preventive Services

he systematic reviews in this report represent the workf the independent, nonfederal Task Force on Com-unity Preventive Services (the Task Force). The Task

orce is developing the Guide to Community Preventiveervices (the Community Guide) with the support of the.S. Department of Health and Human Services in

ollaboration with public and private partners. Theenters for Disease Control and Prevention (CDC)rovide staff support to the Task Force for develop-ent of the Community Guide. A special supplement to

he American Journal of Preventive Medicine, “Introducinghe Guide to Community Preventive Services: Methods,irst Recommendations and Expert Commentary,”ublished in January 2000,23 presents the backgroundnd the methods used in developing the Communityuide. This review of the effectiveness of home visita-

ion on the prevention of violence is one of a series ofommunity Guide reviews on the prevention of violence

hat focus on violence by and against juveniles.

ealthy People 2010 Goals and Objectives

he intervention reviewed here may be useful in reach-ng several objectives specified in Healthy People 2010,24

he disease prevention and health promotion agendaor the United States. These objectives identify some ofhe significant preventable threats to health and focushe efforts of public health systems, legislators, and lawnforcement officials for addressing those threats.any of the proposed Healthy People objectives in Chap-

er 15, “Injury and Violence Prevention,” relate to theome visitation intervention and its proposed effectsn violence-related outcomes. Violence-specific objec-ives that might be related to home visitation are listedn Table 1. (It should be noted, however, that homeisitation can affect other outcomes not directly relatedo violence prevention. As noted, these outcomes areot systematically reviewed here, and the corre-ponding goals and objectives are not included in
able 1.) i
nformation from Other Advisory Groups

n 1991, the U.S. Advisory Board on Child Abuse andeglect, created by Congress, recommended universalome visitation to address maltreatment in the Unitedtates,25 but its recommendation was not accepted byhe U.S. Department of Health and Human Services or

able 1. Selected Healthy People 2010a objectives related toome visitation programs

njury preventionReduce hospitalization for nonfatal head injuries from

60.6 to 45.0 per 100,000 populationb (Objective 15-1).Reduce hospitalization for nonfatal spinal cord injuries

from 4.5 to 2.4 per 100,000 populationa (Objective 15-2).

Reduce nonfatal poisonings from 348.4 to 292 per100,000 populationc (Objective 15-7).

Reduce deaths caused by poisoning from 6.8 to 1.5 per100,000 populationb (Objective 15-8).

Reduce deaths caused by suffocation from 4.1 to 3.0 per100,000 populationb (Objective 15-9).

Reduce hospital emergency department visits from 131 to126 per 1000 populationc (Objective 15-12). See query2.

nintentional injury preventionReduce deaths caused by unintentional injuries from 35.0

to 17.5 per 100,000 populationb (Objective 15-13).(Developmental) Reduce nonfatal unintentional injuries

(Objective 15-14).Reduce drownings from 1.6 to 0.9 per 100,000

populationb (Objective 15-29).iolence and abuse preventionReduce homicides from 6.5 to 3.0 per 100,000

populationb (Objective 15-32).Reduce maltreatment of children from 12.9 (in 1998) to

10.3 per 1000 children aged �18 years (Objective 15-33a).d

Reduce child maltreatment fatalities from 1.6 (in 1998)to 1.4 per 100,000 children aged �18 years (Objective15-33b).d

Reduce the rate of physical assault by current or formerintimate partners from 4.4 (in 1998) to 3.3 per 1000persons aged �12 years (Objective 15-34).

Reduce the annual rate of rape or attempted rape from0.8 (in 1998) to 0.7 per 1000 persons aged �12 years(Objective 15-35).

Reduce sexual assault other than rape from 0.6 (in 1998)to 0.4 per 1000 persons aged �12 years (Objective 15-36).

Reduce physical assaults from 31.1 to 13.6 per 1000persons aged �12 years (Objective 15-37).

Reduce physical fighting among adolescents from 36% to32% (baseline students in grades 9 through 12,fighting during the previous 12 months in 1999)(Objective 15-38).

U.S. Department of Health and Human Services.24

Baseline: 1998 data, age adjusted to year 2000 standard population.Baseline: 1997 data, age adjusted to year 2000 standard population.Note that Objective 15-33a is per 1000 children aged �18 years,hereas Objective 15-33b is per 100,000 children aged �18 years.omparable objectives would be reduction of child maltreatment to290 per 100,000 children aged �18 years and reduction of childaltreatment fatalities to 1.6 per 100,000.

mplemented by Congress.

Am J Prev Med 2005;28(2S1) 13

fYmhon

rcoahds

CgophpRvtDm

Htdeut

C

Tfecsv

trtiaoopa

air

rvo

M

Ieddtbtomdori

ta

F

D

SSA

A

T

C

I

Td

Ftwr

1

Other reviews have found home visitation effectiveor preventing youth violence. The recent report titledouth Violence by the Surgeon General,26 using severaleasures of violent outcomes, concludes that nurseome visitation “has shown significant long-term effectsn violence, delinquency, and related risk factors in aumber of studies.”The Center for the Study and Prevention of Violence

ecommends nurse home visitation for preventinghild abuse and neglect and child violence, amongther benefits. It cites the program designed by Olds etl.27 as a model “blueprint” program that meets itsighest standards of evidence in terms of experimentalesign, substantial effect, replication, andustainability.

The Office of Justice Program’s review, Preventingrime: What Works, What Doesn’t, What’s Promising,28 alsoives a high rating to early home visitation by nurses,ther professionals, and trained paraprofessionals forreventing crime and its risk factors. The CDC cites theome visitation approach among the best practices forreventing youth violence.29 Similarly, Developmentalesearch and Programs, Inc. cites several early homeisitation programs30 (including the nurse home visita-ion program by Olds et al.31 and the Syracuse Familyevelopment Research Program6) among its recom-ended preventive strategies.Finally, the Canadian Task Force on Preventiveealth Care recommends early childhood home visita-

ion programs for preventing child maltreatment inisadvantaged families.32 It notes that the strongestvidence exists for the nurse-delivered programs (assed in the program by Olds et al.27,31) that start beforehe child is born and continue for 2 years after birth.

onceptual Approach and Analytic Framework

he general methods for conducting systematic reviewsor the Community Guide have been described in detaillsewhere.33–37 This section briefly describes the con-eptual approach and determination of outcomes con-idered in assessing the effects of home visitation oniolence.

The conceptual model (or analytic framework) usedo evaluate the effectiveness of home visitation ineducing violence (Figure 1) shows the relationship ofhe intervention to the intermediate outcomes (i.e., thenfluences on parental resources, parenting behavior,nd child development) and finally to the violentutcomes. In this model, we note four broad violentutcome categories (violence by child, violence byarent, intimate partner violence, child maltreatment),ll of which were considered in our review.

Unfortunately, no studies of home visitation reportge-specific effects for either violence by parents orntimate partner violence against parents. Thus, we
eviewed study results for parental violence without r

egard to age, noting that a substantial proportion ofisited parents were themselves adolescents at the timef home visitation program delivery.

ethods

n the Community Guide, evidence is summarized on (1) theffectiveness of interventions; (2) the applicability of evi-ence data (i.e., the extent to which available effectivenessata might apply to diverse population segments and set-ings); (3) positive or negative effects of the interventioneyond those assessed for the purpose of determining effec-iveness, including positive or negative health and nonhealthutcomes; (4) economic impact; and (5) barriers to imple-entation of interventions. When evidence is insufficient to

etermine the effectiveness of the intervention on a specificutcome, information about applicability, economics, or bar-iers to implementation is not included, unless there is anssue of particular interest.

The process used to review evidence systematically andranslate that evidence into the conclusions reached in thisrticle involved:

orming a systematic review development team and a team ofconsultants (see acknowledgments at the end of thisarticle)eveloping a conceptual approach to organizing, grouping,and selecting interventions

electing interventions to evaluateearching for and retrieving evidencessessing the quality of and abstracting information fromeach study

ssessing the quality of and drawing conclusions about thebody of evidence of effectiveness

ranslating the evidence of effectiveness intorecommendations

onsidering data on applicability, other effects, economicimpact, and barriers to implementation

dentifying and summarizing research gaps

his section summarizes how these methods were used ineveloping the reviews of home visitation interventions. The

Home

visitation

program

Violence by visited

parent(s) (other than child

maltreatment)

Intimate partner

violence against the visited parent

Violence against

child (maltreatment)

Violence by visited

child

Parent

Knowledge

SkillsSelf-confidence

Access to resources

Parenting

Child

Development

Skills

Health/well-being

igure 1. Analytic framework for early childhood home visi-ation. Circle represents the intervention itself; rectanglesith rounded corners show intermediate outcomes; andectangles with square corners show health-related outcomes.

eviews were produced by the systematic review development

ber 2S1

ta

S

EEvCrctrss

r

EPACPC

ip(pspccbti

A

Erirmalei

A

Eutsvseaecc

Ao

Turtwsrtlf

c

w

I

I

C

C

c

c

dcuiiiazir

e

a

AIGINMU

eam and a multidisciplinary team of specialists and consult-nts representing a variety of perspectives on violence.

earch for Evidence

lectronic searches for literature were conducted in Medline,MBASE, ERIC, NTIS (National Technical Information Ser-ice), PsycINFO, Sociological Abstracts, NCJRS (Nationalriminal Justice Reference Service), and CINAHL. We also

eviewed the references listed in all retrieved articles, andonsulted with experts on the systematic review developmenteam and elsewhere. We used journal papers, governmenteports, books, and book chapters. The initial literatureearch on the topic was conducted in August 2000 and aecond (update) search was conducted in July 2001.

Articles were considered for inclusion in the systematiceview if they had the following characteristics:

valuated the specified intervention.ublished before July 2001.ssessed at least one of the violent outcomes specified.onducted in an established market economy.a

rimary study rather than, for example, a guideline or review.ompared a group of people who had been exposed to theintervention with a group of people who had not beenexposed or who had been less exposed. (The comparisonscould be concurrent or in the same group over time.)

The four outcomes evaluated to determine the effect of thentervention were violence by the child, violence by thearent, intimate partner violence, and child maltreatmentabuse and neglect). Specific measures accepted as direct orroxy measures of these outcomes are listed in the “Results”ection. If both direct and proxy measures were available,reference was given to the direct measure. The four out-omes reviewed are referred to as “recommendation out-omes,” because, if evidence is sufficient, they provide theasis for recommending the intervention (i.e., we acceptedhese as either representing improved health or as proxies formproved health outcomes).

bstraction and Evaluation of Studies

ach study that met the inclusion criteria was read by twoeviewers who used a standardized abstraction form to recordnformation from the study.37 Any disagreements between theeviewers were reconciled by consensus among the develop-ent team members. In addition, to ensure a consistent

pplication of both assessments of study design suitability andimitations in execution quality within the body of evidence,very evaluated study was presented and discussed in meet-ngs of the systematic review development team.

Established market economies as defined by the World Bank includendorra, Australia, Austria, Belgium, Bermuda, Canada, Channel

slands, Denmark, Faeroe Islands, Finland, France, Germany,ibraltar, Greece, Greenland, Holy See, Iceland, Ireland, Isle of Man,

taly, Japan, Liechtenstein, Luxembourg, Monaco, the Netherlands,ew Zealand, Norway, Portugal, San Marino, Spain, St. Pierre and

Wiquelon, Sweden, Switzerland, the United Kingdom, and thenited States.

ssessing the Suitability of Study Design

ach study that met the inclusion criteria was evaluated bysing a standardized abstraction form (available at www.

hecommunityguide.org/methods/abstractionform.pdf) and was as-essed for suitability of the study design and threats toalidity.33 On the basis of the number of threats to validity,tudies were characterized as having good, fair, or limitedxecution.33,37 Studies with good or fair quality of executionnd any level of design suitability were included in the body ofvidence. Our study design classifications, chosen to ensureonsistency in the review process, sometimes differ from thelassification or nomenclature used in the original studies.

ssessing the Quality and Summarizing the Bodyf Evidence on Effectiveness

he quality of study execution was systematically assessedsing Community Guide methods.33 Several studies had sepa-ate intervention “arms,” that is, two or more interventionshat were compared with each other or a control; each armas assessed separately. Unless otherwise noted, we repre-

ented the results of each study arm as a point estimate of theelative change in the violent outcome rate associated withhe intervention. We calculated percent point changes (abso-ute percent change) and baselines using the followingormulas:

For studies with before-and-after measurements and con-urrent comparison groups:

Effect size � (Ipost ⁄ Ipre) ⁄ (Cpost ⁄ Cpre) � 1

here

post � last reported outcome rate in the intervention groupafter the intervention

pre � reported outcome rate in the intervention groupbefore the intervention

post � last reported outcome rate in the comparison groupafter the intervention

pre � reported outcome rate in the comparison groupbefore the intervention

For studies with post measurements only and concurrentomparison groups:

Effect size � (Ipost � Cpost) ⁄ Cpost

For studies with before-and-after measurements but nooncurrent comparison:

Effect size � (Ipost � Ipre) ⁄ Ipre

We report the effect of the intervention as beneficial oresirable when the intervention was associated with a de-rease in a violent outcome examined, and as harmful orndesirable when the intervention was associated with an

ncrease in the violent outcome. We use the median andnterquartile range to report effect sizes from multiple stud-es. We also note whether zero was included within the uppernd lower interquartile range. Interquartile ranges includingero suggest that the results are inconsistent in direction;nterquartile ranges not including zero suggest that theesults are consistent in direction.

In some cases, we had to select among several possibleffect measures for our summary measures of effectiveness.
hen available, we included measures adjusted for potential
Am J Prev Med 2005;28(2S1) 15

cmsiffavs

ttrbdesoTepccSoWqcg

ioen

S

TunTdl

O

Anoaasrvns

E

Etoe

SI

Bwi

S

Siioirattatqeo

REE

WhlvvDasto(oorbt

Eiptapqqs

eyt

1

onfounders in multivariate analysis rather than crude effecteasures. No studies were excluded from the evaluation

trictly on the basis of an insufficient follow-up period. If thentervention program had multiple evaluations at differentollow-up points, we chose the evaluation at the longestollow-up period with an attrition rate of �30%. In thenalysis of study findings, we used the standard two-tailed palue cut-off at the 0.05 level as a measure of statisticalignificance.

We summarized the strength of the body of evidence onhe basis of the number of available studies, the strength ofheir design and execution, and the size and consistency ofeported effects, as described in detail elsewhere.33 In brief,y Community Guide standards, single studies of greatestesign suitability and good execution can provide sufficientvidence of effectiveness if the effect size is itself consideredufficient; single studies are generally considered sufficientnly if the effect measure is statistically significant (p �0.05).hree studies of at least moderate design suitability and fairxecution, or five studies with at least fair execution, canrovide sufficient evidence of effectiveness if the findings areonsistent in direction and size and if the effect size is itselfonsidered sufficient (i.e., of public health importance).tatistical significance is considered principally when there isnly one study of greatest suitability and good execution.hen the number of studies and their design and execution

uality are sufficient, by Community Guide standards, to draw aonclusion on effectiveness, the results are summarized bothraphically and statistically.It is critical to note that when we conclude that evidence is

nsufficient to determine the effectiveness of the interventionn a given outcome, we mean that we do not yet know whatffect, if any, the intervention has on that outcome. We doot mean that the intervention has no effect on the outcome.

ummarizing Applicability

he body of evidence used to assess effectiveness was alsosed to assess applicability, the generalizability of effective-ess findings to populations with differing characteristics.he systematic review development team and the Task Forcerew conclusions about the applicability of the available

iterature to various populations and circumstances.

ther Effects

s noted, the Community Guide review of home visitation didot systematically assess the effects of this intervention onther outcomes (e.g., on mother–infant attachment, physicalnd cognitive development, school achievement, substancebuse, or other behavior problems). However, we mentionome of the benefits noted in the studies that we haveeviewed. We also note the potential harms of the homeisitation intervention if they were mentioned in the effective-ess literature or were thought to be of importance by theystematic review development team.

conomic Evaluations

conomic evaluations of interventions were conducted only ifhere was sufficient or strong evidence of effectiveness. Meth-ds used in economic evaluations are described
lsewhere.35,36 d

ummarizing Barriers to Implementation ofnterventions

arriers to implementation were summarized only if thereas sufficient or strong evidence of effectiveness of the

ntervention.

ummarizing Research Gaps

ystematic reviews in the Community Guide identify existingnformation on which to base public health decisions aboutmplementing interventions. An important additional benefitf these reviews is identification of areas in which information

s lacking or of poor quality. To summarize these deficits,emaining research questions for each intervention evaluatedre identified. Where evidence of effectiveness of an interven-ion is sufficient or strong, remaining questions about effec-iveness, applicability, other effects, economic consequences,nd barriers are summarized. Where evidence is insufficiento determine effectiveness of an intervention, remaininguestions are summarized only about effectiveness and otherffects, but not about applicability, economic consequences,r barriers.

esults: Part I—Intervention Effectiveness andconomic Efficiencyvidence Reviews: Violence by Child

e reviewed the evidence concerning the violent be-avior of children who were home-visited early in their

ives. Although the prevention of youth suicide (i.e.,iolence against self) is a plausible outcome of homeisitation, we found no study that assessed this outcome.irect measures for violence by child were reported

nd observed violence, and violent crime; proxy mea-ures were arrests, convictions, or delinquency as ascer-ained from official records (all for behavior that mightr might not include violence), externalizing behaviorbehavior in which psychological problems are actedut), and conduct disorder (in which “the basic rightsf others or major age-appropriate societal norms orules are violated”).38 Studies were included in theody of evidence only if they reported at least one ofhese effect measures.

ffectiveness. Our search identified four stud-es6,31,39,40 that reported the effects of home visitationrograms on violence by the visited children. Descrip-ive information about execution quality, design suit-bility, and outcomes evaluated in these studies isrovided in Appendix A. Two studies31,39 were of gooduality of execution, and two studies6,40 were of fairuality. All four studies were of highest designuitability.

One study31 of nurse home visitation in Elmira NYxamined criminal and delinquency outcomes in 15-ear-olds who were visited by nurses prenatally throughhe first 2 years of their lives. The systematic review
evelopment team chose self-reported delinquency
ber 2S1

(cbcdivT1iTc(rtsnwcm

pl17ci(Atms5cg

iCfaBts

Cchcibnm

ftBeov

E

Wvvbptrcpeo

Esttvsi

t6stcsrar8

CaavcsTmitws

E

Wvc

b

fsbs

i.e., “major delinquent acts”)b as the principal out-ome in this study because it referred to self-reportedehavior, unaffected by the social processes of arrest oronviction. Three of the items included as “majorelinquent acts” are explicitly violent, and the remain-

ng items have violent connotations (i.e., the threat ofiolence or the violation of property and its owners).he study reported a nonsignificant increase (of8.2%) in self-reported major delinquent acts in thentervention group compared with the control group.he study also reported a statistically significant de-rease in self-reported arrests (58.2%) and convictions63.0%), and nonsignificant decrease (33.3%) in ar-ests of subjects reported by subjects’ mothers. Amonghe children of single mothers of low socioeconomictatus (SES), home visitation was associated with aonsignificant reduction in major delinquent acts andith significant reductions in self-reported arrests andonvictions, as well as arrests reported by the child’sother.Another study6 of a multicomponent home visitation

rogram in Syracuse NY assessed delinquent and vio-ent outcomes when visited children had reached 13 to6 years. Researchers reported a significantly lower (by2.3%) proportion of subjects “processed as probationases by the County Probation Department” in thentervention group compared with the control groupprobation processing is an indicator of serious crime).

listing of offenses by all study subjects indicates thathe offenses committed by comparison subjects were

ore serious than those committed by home-visitedubjects. In addition, the study reported that 2 (out of4) subjects in the control group committed violentrimes, whereas none of 65 subjects in the interventionroup committed such crimes.The other two studies39,40 reported only externaliz-

ng behavior (from the Externalizing subscale of thehild Behavior Checklist41). One study39 reported such

ollow-up results when the children were aged 9 years,nd the other study40 when children were aged 5 years.oth studies reported no significant differences be-

ween intervention and control groups in Externalizingubscale scores.

onclusion. Although the number of studies is suffi-ient to draw a conclusion about the effectiveness ofome visitation in preventing later violence by visitedhildren, study findings are inconsistent. Two stud-es39,40 found no significant differences in outcomesetween intervention and control populations (but didot report results in a manner that allowed an assess-ent of the direction of the study findings), one study6

“Major delinquent acts” were defined in the study as any of theollowing: hurt someone sufficiently that they needed bandages, stoleomething, trespassed, damaged property on purpose, hit someone

wecause he or she said something objectionable, carried a weapon,et fire intentionally, or was in a fight with gang members.

ound a beneficial effect, and one31 had mixed resultshat included benefits as well as nonsignificant effects.ecause the findings from these studies are mixed, thevidence is insufficient to determine the effectivenessf home visitation interventions in preventing childiolence.

vidence Reviews: Violence by Parents

e reviewed the evidence on the effect of homeisitation on the violent behavior of parents in theisited home (other than maltreatment, consideredelow). Direct measures for this outcome were re-orted and observed violence, and arrests or convic-ions for violent crime (from self-reports or officialeports). Proxy measures were general arrests andonvictions, which did not state whether violence wasart of the crime. Studies were included in the body ofvidence only if they reported at least one of theseutcome measures.

ffectiveness. The 15-year follow-up to the Elmiratudy42 (of highest design suitability and good execu-ion quality) was the only study identified by our searchhat examined the effect of home visitation on parentaliolence. Many of the mothers (48% of the initialample) were teenagers when home visitation wasnitiated.

This study reported statistically nonsignificant reduc-ions in maternal arrests (59.1% by self-report and8.4% by state records) and convictions (76.9% byelf-report and 55.6% by state records) for mothers inhe intervention group compared with mothers in theontrol group. In the subsample of mothers who wereingle and of low SES at the time of visitation, the studyeported statistically significant reductions in maternalrrests (69.0% by self-report and 82.2% by stateecords) and convictions (78.6% by self-report and1.2% by state records).

onclusion. The one study of highest design suitabilitynd good execution quality could be sufficient to allowssessment of the effectiveness of home visitation inter-entions in preventing parental violence (other thanhild maltreatment). However, because of the lack oftatistically significant results for the total sample, theask Force judged the evidence insufficient to deter-ine the effectiveness of home visitation interventions

n preventing parental violence. The finding of substan-ial (although not statistically significant) effects in thehole sample, and statistically significant effects in a

ubsample, are promising and deserve replication.

vidence Reviews: Intimate Partner Violence

e reviewed the evidence on the effect of homeisitation on violence involving the parents of visitedhildren. Direct measures for this outcome category
ere reported and observed partner victimization, as
Am J Prev Med 2005;28(2S1) 17

wpio

Esttpeuwtvsvwr

Caqlmp

E

Wtgospwipftiafi

Ep(fTwOwtieu

h

pttmoamttaoi

dotietio(a(poF

aevTammrcphvmta(tromrsa�rote

1

ell as arrests and convictions for partner assault. Noroxy measures were considered. Studies were included

n the body of evidence only if they reported at leastne of these outcome measures.

ffectiveness. The 15-year follow-up to the Elmiratudy43 (of highest design suitability and good execu-ion quality) was the only study identified by our searchhat examined the effect of home visitation on intimateartner violence. Descriptive information about thexecution quality, design suitability, and outcome eval-ated in this study is provided in Appendix A. Among aide range of outcomes examined, the study included

he incidence of domestic violence in the families ofisited children over the 15-year follow-up period. Noignificant difference in the incidence of domesticiolence between the intervention and control groupsas found, and no effect size or direction of effect waseported.

onclusion. Although the one study was characterizeds having highest design suitability and good executionuality, its failure to find a statistically significant effect

eads to a conclusion of insufficient evidence to deter-ine the effectiveness of home visitation in preventing

arental violence.

vidence Reviews: Child Maltreatment

e reviewed evidence about the effects of home visita-ion on the subsequent maltreatment (abuse or ne-lect) of visited children. Direct measures for thisutcome category were reports from child protectiveervices, and abuse or neglect reported or observed byarents or others. Proxy measures for this categoryere emergency room visits or hospitalizations for

njury or ingestion, reported injury, and out-of-homelacement. Although these proxy outcomes may resultrom causes other than abuse or neglect, many arehought to result from these causes.44 Studies werencluded in the body of evidence only if they reportedt least one of these outcome measures. For otherorms of child victimization, such as bullying, no qual-fying studies were identified.

ffectiveness. We identified 22 studies (in 21 re-orts)42,45–64 with 27 intervention arms. One study45

representing one intervention arm) was excludedrom the review because of limited quality of execution.he remaining 21 studies (with 26 intervention arms)ere included in the body of evidence for this review.ne report64 described two separate studies, one ofhich had two intervention arms. One study62 had

hree intervention arms, and two studies55,59 had twontervention arms each. Descriptive information aboutxecution quality, design suitability, and outcomes eval-ated in these studies is provided in Appendix A.Twenty intervention arms assessed the effects of

ome visitation on abuse or neglect (reported by child m


rotective services or by home visitors); five interven-ion arms assessed effects on rates of injury, trauma, orhe ingestion of poison (from emergency room visits,

edical or hospital records, or mothers’ reports); andne intervention arm assessed out-of-home placements an outcome (see Appendix A). Most studies assessedaltreatment, injury, or trauma at the conclusion of

he intervention; 10 months was the shortest follow-upime, and 3 years the longest. Only one study42 assessedbuse and neglect substantially beyond the conclusionf the intervention: 15 years after the intervention’s

nitiation (i.e., when the children were aged 15 years).Overall, summary effect measures were in the desired

irection (i.e., the intervention group had a lower ratef undesirable outcomes—abuse or neglect, injury orrauma, or out-of-home placement—than the compar-son group) in 19 of the 26 study arms in the body ofvidence; effect measures in the remaining 7 interven-ion arms were not in the desired direction (i.e., thentervention group had a higher rate of undesirableutcomes). Overall, the median effect size was �38.9%interquartile range, �74.1% to �24.0%). Results ofnalyses stratified by the type of the outcome measureabuse or neglect, injury or trauma, or out-of-homelacement) are presented in Table 2. The distributionf effect sizes stratified by outcome is presented inigure 2.Researchers65 have noted that, because home visitors

re legally required to report maltreatment, the pres-nce of the home visitor increases the likelihood thatiolence (such as child maltreatment) will be observed.hus, the presence of the visitor biases all of the studiesgainst the hypothesis that visitation prevents maltreat-ent. Two studies in this review47,50 allow the assess-ent of the magnitude of this bias because they report

ates of maltreatment assessed among home-visitedhildren by child protective services alone and by childrotective services in conjunction with reports by theome visitor. In these two studies, the presence of theisitor seemed to increase the rate of reported maltreat-ent by 80%47 and 150%,50 respectively. We explored

he implications of such a bias in a sensitivity analysis bydjusting results using a more conservative estimatei.e., an increase of 50%) to assess the possible effects ofhis bias in the studies we reviewed. These adjustedesults (for the abuse or neglect outcome only, adjustedn the assumption that home visitors would report 50%ore maltreatment cases) are presented as a separate

ow in Table 2 (“Abuse/neglect adjusted”). As can beeen, after adjustment the median effect size for thebuse or neglect outcome changes from �39.6% to59.7%, and the upper boundary of the interquartile

ange moves below zero. Although the exact magnitudef this bias is open to debate, overall the presence ofhis reporting bias would tend to strengthen the gen-ral conclusion that home visiting reduces child
altreatment.
ber 2S1

mssobotc

ct

tmmasowfpe(nFd

nasmte[ipgih

T

TTB

B

B

R

B

a

b

N

F

We found that professional visitors (i.e., nurses andental health workers) produced more beneficial re-

ults than paraprofessionals (Table 2). (Most studiesimply reported using “paraprofessional” visitors with-ut defining the term; we assumed paraprofessionals toe trained for the visiting program, but to lack formalr professional training in subjects relevant to visita-ion, such as health care, mental health, or familyounseling.)

In further analyses, we also found strong negativeorrelation (Spearman rho��0.52; p �0.01) betweenhe observed effect size and the planned duration of

able 2. Summary analysis of studies measuring child maltre

Studies(n)

Negative(n)

P

otal papers 21otal intervention arms 26 19 7y outcomeAbuse/neglect 20 14 6Abuse/neglect (adjusted)b 20 17 3Injury/ingestion/trauma 5 5 0Out-of-home placement 1 0 1

y visitor typeNurses 5 5 0Paraprofessionals 18 11 7Mental health workers 3 3 0

y componentsSingle component 17 13 4Multicomponent 9 6 3

andomizationRandomized 18 13 5Nonrandomized 8 6 2

y time of program initiationPrenatal 6 4 2Postnatal 16 11 5Both prenatal andpostnatal

4 4 0

Percentage change, intervention group versus control (see EffectiveAdjusted for detection bias (see text), by a factor of 1.5.A, not available or not applicable.

migure 2. Distribution of effect sizes stratified by outcome.

he program (i.e., longer programs tended to showore beneficial results). (Information on the actual orean number of home visits made was not available insufficient number of studies reviewed to allow analy-

is; thus we analyzed only intended number or durationf visits.) The correlation was particularly strong whene considered only the programs delivered by parapro-

essionals (Spearman rho��0.63; p �0.01). For para-rofessional visitors, effects are mixed, and beneficialffects are found only in programs of longer durationi.e., two years or longer). We found few studies ofurse visitation programs with a duration of �2 years.igure 3 shows the distribution of study effect sizesepending on program duration and the type of visitor.The other stratified analyses that we performed did

ot show any substantial or consistent differencescross strata (Table 2). Studies in these analyses weretratified by method of subjects’ assignment to treat-ent conditions (randomized vs nonrandomized), the

ime of program initiation (prenatal vs postnatal vsither), and program components (single-componenti.e., home visitation only] vs multicomponent [i.e.,nvolving some additional services, such as child care,ediatric care, free transportation, or parent supportroups]). Available studies did not provide enoughnformation to determine whether effectiveness ofome visitation differs for first-time mothers versus

nt (abuse and neglect)

eMediana

Interquartilerangea Rangea

Lower Upper Lower Upper

�38.9 �74.1 24 �100 228.4

�39.6 �74.6 37.2 �100 228.4�59.7 �83.1 �8.5 �100 118.9�31.9 �72.2 �10.8 �100 �2.9

13.0 NA NA NA NA

�48.7 �89.0 �24.6 �100 �2.9�17.7 �65.7 41.2 �100 228.4�44.5 NA NA �93.2 �18.7

�31.9 �55.9 5.5 �87.4 127.6�46.3 �96.6 35.1 �100 228.4

�27.5 �46.9 19.6 �100 228.4�68.3 �77.3 20.0 �93.2 94.1

�23.7 �58.0 66.8 �93.2 228.4�20.9 �59.5 37.2 �100 127.6�74.3 NA NA NA NA

ection in Evidence Reviews: Child Maltreatment).

atme

ositiv(n)

ness s

others with previous live births.

Am J Prev Med 2005;28(2S1) 19

Aeicejfopmpwia

tanvmpast

Ooiotsn1(mdf2uit

ced(pipiAvipaso

tlcncrsao

vmss

Enrrsvtwbabvcsfchma

l

c

d

FSps

2

pplicability. The same body of evidence was used tovaluate the applicability of home visitation programsn different settings and populations. All studies wereonducted in the continental United States and Hawaii,xcept for one in Canada (Montreal).59 The vast ma-ority of programs targeted those populations andamilies believed to benefit most from componentsffered by many of the programs reviewed (e.g., sup-ort in parenting and life skills, prenatal care, and caseanagement). Target populations included teenage

arents; single mothers; families of low SES; familiesith very low birth-weight infants; parents previously

nvestigated for child maltreatment; and parents withlcohol, drug, or mental health problems.

An analysis of the effects of the Elmira home visita-ion program on intimate partner violence66 indicatesn inverse relationship between the frequency of part-er violence in a home and the effectiveness of homeisitation programs, suggesting that partner violenceay need to be addressed before home visitation

rograms can be effective in reducing violence directedgainst children in the home. No study reviewed as-essed the effectiveness of universal home visitation inhe prevention of violence.

ther positive or negative effects. Systematic analysisf the many other possible beneficial or harmful effects

s beyond the scope of this review. We found examplesf such effects in the best study to date, which also hadhe longest follow-up period (15 years).31,42 Thattudy42 reported consistently beneficial, but statisticallyonsignificant, effects for visited mothers, including a9.0% decrease in number of subsequent pregnanciesa risk factor for child abuse),10 a 19.9% decrease inonths receiving Aid to Families with Dependent Chil-

ren (AFDC), a 15.1% decrease in months receivingood stamps, a 7.5% increase in time employed, and a0.9% decrease in problems related to illicit substancese. Results for these outcomes were statistically signif-

cant in the study subsample (40% of the total sample)

-120

-60

0

60

120

180

240

0 0.5 1 1.5 2 2.5 3 3.5 4

Duration of the program, years

Rela

tive p

erc

ent

change

Paraprofessionals Professionals

igure 3. Home visitation effects by duration of program.pearman correlation: whole sample: �0.52 (p �0.01); para-rofessionals: �0.63 (p �0.01); professionals: �0.32 (notignificant).

hat included only single, low SES mothers.pb


Consequences other than violence for the visitedhildren at 15-year follow-up31 were less clear. Forxample, this study found decreases in the incidence ofrug use (12.3%), in the number of sexual partners25.6%), and in the number of long-term school sus-ensions (75%). At the same time it reported increases

n the incidence of alcohol use (19.1%) and in theroportion of subjects who ever had sex (20.0%) in the

ntervention group compared with the control group.mong the children of low SES single mothers, homeisitation was generally associated with desired results,ncluding a significant reduction in the number of sexartners; nonsignificant reductions in the use of drugs,lcohol, and cigarettes; an increase in the number ofhort-term suspensions; but a decrease in the numberf long-term suspensions.Other possible beneficial effects of the home visita-

ion programs reviewed that are mentioned in theiterature include improved socioemotional and physi-al development of the visited children; better immu-ization coverage; better access to, and use of, medicalare; improved family planning; improved home envi-onment; and a higher level of education and profes-ional achievement attained by the parents.1,2 However,ssessment of these other outcomes is beyond the scopef the present review.It has been suggested that negative effects of home

isitation (especially when targeted to high-risk groups)ay include stigmatization of the target group (e.g.,

ingle mothers, minority, poor), but we found notudies of this issue.28,42

conomic efficiency. Our search identified one eco-omic evaluation of a home visitation intervention toeduce child abuse and neglect. This study65 was car-ied out in a semirural county in upstate New York. Thetudy evaluated the net benefits of a nurse homeisitation program provided to first-time mothers. Ofhe mothers in the study, 61% were of low SES and 24%ere either unmarried or aged �19 years. Home visitsy a registered nurse began before the child was bornnd lasted until the child reached age 2 years. The visitsegan on a weekly basis; by 20 months postdelivery,isits were made every 6 weeks. Program content in-luded parent education, the strengthening of familyupport (by encouraging other family members andriends to become involved in the home visit and inhild care), and the linking of families with otherealth and human services. Goals included improve-ent of the child’s health, reduction of child abuse,

nd improvement of the mother’s own life course.The cost-benefit analysisc of this intervention was

imited to government costs and benefits, not those of

Cost–benefit analysis is an evaluation technique that standardizes, inollar terms, both the costs and benefits accrued in a given time
eriod. Results are typically reported as a single value (e.g., netenefits [total benefit minus total cost]).
ber 2S1

pacapWpbbssn

ilo$lnblFbftttdmtr

qGp

dts11eecsfsaiaawncp

i

diwawawvecitmcs�rdialweepvaeccelae

BidrHfsfoaabpatbmn

Cet

rogram participants, the healthcare system, or societyt large. Program costs considered were through thehild’s second birthday, and included nurse salariesnd fringe benefits; nurse training; part-time secretary;art-time supervisor; taxicab; linked services such as theomen, Infants, and Children (WIC) nutritional sup-

lementation program; supplies; and overhead. Theenefits considered were through the child’s fourthirthday, and included reduced use of governmentervices (i.e., AFDC, child protective services, foodtamps, and Medicaid), and newly generated tax reve-ues from mothers returning to work.Authors reported results for the subsample of low-

ncome mothers as well as for the whole sample. For theow-income subsample, government benefits more thanffset program costs, for a net benefit to government of350.61 per low-income family (adjusted to 1997 dol-ars). Including benefits attributable only to reducedeed for child maltreatment services (3% of totalenefits) was not enough to offset program costs in the

ow-income subsample (i.e., costs exceeded benefits).or the whole sample, government costs exceededenefits, which resulted in a net benefit of �$3081 peramily (adjusted to 1997 dollars). Benefits attributableo reduced child maltreatment were not specified forhe whole sample. Including benefits beyond those ofhe government, such as averted healthcare costs, pro-uctivity losses, and other possible monetary and non-onetary benefits67 associated with reduced child mal-

reatment, or any longer-term benefits, would likelyesult in greater net benefits.

The study was classified as satisfactory, based on theuality assessment criteria used in the Communityuide.35 The economic summary table for the study isrovided at the website (www.thecommunityguide.org).36

For the above program, adjusted nurse visitationirect costs—including salaries, fringe benefits, part-ime supervisor and secretary, overhead, travel, andupplies—were estimated to be $6286 per family in997 dollars over the 2-year intervention period. In a998 follow-up investigation,27 program costs were re-stimated to be $7000 per family (in 1997 dollars). Thisstimate was based on the original study design, but wasalculated to serve 100 families with four full-time nursepecialists, each taking on �25 cases. In addition to theull-time nurses, the new estimate includes a part-timeecretary and nurse supervisor; comprehensive officend program materials, including cell phones; liabilitynsurance; medical supplies; general staff development;nd mileage. In most cases, training and technicalssistance, including a computer and network fees,ould also be necessary at program outset (but wereot included in the base case analysis). Such startuposts were estimated to increase program costs to $8000er family during the first 3 years of the program.Another study48 with less-intensive and less-frequent

ntervention (i.e., five visits over 18 months) was con- e

ucted at the Hospital of the University of Pennsylvanian Philadelphia. Early discharge and home visitationere only carried out if the infant’s physical conditionnd environment met specified criteria (e.g., clinicallyell, stable maintenance of body temperature, anddequate home care facilities), and if parental consentas given. Program costs of nurse home visitation forery low birth-weight infants that had been dischargedarly were estimated and included pre- and post-dis-harge nurse time, telephone, and travel expenses. Asn the other study reporting economic characteristics,he population studied here was also of lower SES and

ost mothers were single. The effect of the program,ompared with a nonvisited control group, was alsoimilar to that of the other study, that is, �48.7% versus46.3%, both close to the median of the studies

eviewed. Average program costs (adjusted to 1997ollars) were estimated to be $958 per family. Infants

ncluded in the study were born between October 1982nd December 1984 and received postdischarge fol-ow-up care by either a full-time or part-time specialistith a master’s degree in nursing. Predischarge visitsstablished a relationship between the nurse and par-nts to facilitate training and information exchange torevent abuse and neglect. Postdischarge visits pro-ided further instruction and assessment of both infantnd parent well-being. Nurses also contacted the par-nts by telephone during the first 8 weeks postdis-harge and were on call to address immediate con-erns. The large difference between this program coststimate and that provided by Olds et al.31,42 is mostikely due to program duration and frequency of visitss well as additional program costs included in thestimate.

arriers to intervention implementation. Barriers tomplementing home visitation interventions frequentlyiscussed in the literature include difficulties in theetention of study participants68 and program staff.2,65

ome interventions have generally been targeted toamilies of low SES, who are in challenging life circum-tances with few resources. It is understandable, there-ore, that such families might be overwhelmed withther problems and might lack sustained interest in orbility to commit to regular home visitation; they mightlso be hard to reach and retain in the programecause of frequent life transitions.69,70 Home visitingersonnel (especially when paraprofessional lay visitorsre used) may be hard to recruit, train, and retain dueo low pay and difficult work conditions. It has alsoeen noted that paraprofessional visitors may requireore training and supervision than professionals (e.g.,

urses).71

onclusion. According to Community Guide rules ofvidence,33 available studies provide strong evidencehat early childhood home visitation programs are
ffective in preventing child maltreatment, reducing
Am J Prev Med 2005;28(2S1) 21

rgmtfed

R

Sesgihdmlvt

E

AhmlptvcifoscbFs

hirfiittiofbtste

pt

W

W

W

H

H

W

W

W

S

A

Tmgwtvdivue“o

pdappcetdi(

2

eported maltreatment by approximately 39%. Pro-rams delivered by professional visitors (nurses orental health workers) seem to yield greater effects

han those delivered by paraprofessionals. For parapro-essional visitors, effects are mixed, and beneficialffects are generally found in programs of longeruration (i.e., �2 years).

esults: Part II—Research Issues

trong evidence indicates that early home visitation isffective in preventing child maltreatment in low SES,ingle mother, and other families that have been tar-eted by such programs. Currently available evidence isnsufficient to determine the effectiveness of earlyome visitation in preventing violence by visited chil-ren or visited parents (i.e., violence other than childaltreatment), or in preventing intimate partner vio-

ence in visited families. For all four outcomes re-iewed, we identified key research issues in several areashat have not been answered or merit further research.

ffectiveness

lthough we have demonstrated the effectiveness ofome visitation in the prevention of child maltreat-ent, evidence on the other outcomes assessed (vio-

ence by children, violence by parents, and intimateartner violence) was insufficient to determine effec-iveness. Further research on the effectiveness of homeisitation in the prevention of these outcomes wouldlarify other possible benefits of this intervention. Find-ngs of large, but statistically nonsignificant, effect sizesor some of these outcomes suggest that studies may bef low statistical power; we believe that larger sampleizes should be considered. Suicidal behavior by visitedhildren and diverse forms of victimization should alsoe assessed as outcomes in home visitation programs.ollow-up studies should determine long-term as well ashort-term effects.

The evidence we reviewed indicates a benefit ofome visitation for the reduction of child maltreatment

n populations that have been shown to be at elevatedisk of maltreatment. The population that might bene-t is a large one. In 1999, 33% of the 3.6 million births

n the United States were to single mothers, 12% wereo teen mothers, and 22% were to mothers with lesshan a high school education72; 43% of births—approx-mately 1.7 million—were to mothers with at least onef these characteristics (B. Hamilton, National Centeror Health Statistics, personal communication, Septem-er 9, 2002). Given such a large need, it will be usefulo conduct research, perhaps in the form of demon-tration projects, to make the intervention more effec-ive. Because the visitation programs reviewed are het-
rogeneous and differ in content, organization, e

ersonnel, intensity, and other characteristics, ques-ions that should be addressed include:

hat number, spacing, and duration of home visits isoptimal for cost-effective programs that are accept-able to visited families?hat training for professional and paraprofessionalhome visitors maximizes cost-effectiveness?hat circumstances enhance the effectiveness of para-professional visitors (e.g., educational backgroundand origin)?ow should the curriculum of home visits be orga-nized, in terms of structure, and specific componentsand contents?ow strong is the need for program fidelity (i.e.,degree of adherence to initially proposed curriculumand schedule) for the reduction of violent behaviors?hat is the utility of additional components, such asparent support groups, child daycare, enhanced pe-diatric care, free transportation to appointments,and linkage with social support services?hat are the essential components of home visitationprograms, and what components are dispensable?hat populations are most likely to benefit from homevisitation programs and what program characteristicsare most important for specific populations?

tudies of some of these issues are under way.73

pplicability

he effectiveness of home visitation for child maltreat-ent prevention has been demonstrated in a variety of

eographic areas and “at-risk” populations. Althoughe found insufficient evidence to determine the effec-

iveness of home visitation on child violence, parentaliolence, and other outcomes among both visited chil-ren and parents, evidence from the Elmira study

ndicated beneficial effects for these outcomes amongisited low SES households with single parents. It is stillnclear whether other specific subgroups (e.g., racial/thnic populations) within the general category ofpopulation at-risk” are likely to benefit more thanther subgroups.Studies of the effectiveness of home visitation in

reventing violence by visited children have examinediverse populations, but too few studies are available,nd they provide inconsistent evidence. Evidence aboutarental violence outcomes is limited to a mostly whiteopulation from the northeastern United States, prin-ipally from the study by Olds et al.31,42 If found to beffective, the applicability of early home visitation forhese outcomes in different populations should also beetermined. In addition, it will be useful to determine

f home visitation is effective in the general populationas well as in “at-risk” populations), and if so, if benefits
xceed costs.
ber 2S1

O

Aegiduaepn

hmv

b

H

I

W

E

Te

W

H

H

H

B

SmAt

W

C

W

W

W

W

W

D

Thlfmimbiietovab

gSdtitpggaafiv

tbeCVpdtC

ther Positive or Negative Effects

s noted, this review did not systematically summarizevidence of the effectiveness of home visitation pro-rams on nonviolent outcomes. Such outcomes mightnclude children’s cognitive, emotional, and physicalevelopment; school achievement; substance use; sex-al activity; access to health care; immunization cover-ge; quality and safety of the home environment;mployment of parents; educational achievement ofarents; and family planning, including spacing andumber of subsequent pregnancies.We are hopeful that the research questions that we

ave just developed for home visiting and violenceight also inform additional studies or reviews of home

isiting to achieve other outcomes as well.Concerning negative effects, questions that should

e addressed include:

ow serious is the problem of stigmatization by riskcriteria when home visitation programs are directedat “at-risk” populations?

f stigmatization is an important problem (under someor all circumstances), what can be done in programdesign to minimize the negative effects ofstigmatization?hat role can community coalitions play in preventingor alleviating stigmatization?

conomic Evaluations

he available economic evidence was limited. Consid-rable research is warranted on the following questions:

hat is the cost and cost effectiveness of the variousalternative home visitation programs?ow can effectiveness in terms of health outcomes orquality-adjusted health outcomes be better mea-sured, estimated, or modeled?ow can the cost benefit of this program be estimatedfrom a societal perspective?ow do specific characteristics of this approach con-tribute to economic efficiency?

arriers

everal important barriers may adversely affect imple-entation and outcomes of home visitation programs.ddressing the following research questions may help

o avoid or overcome these barriers:

hat program components or design features improvethe retention of program participants?

an baseline characteristics of families that are morelikely to drop out of home visitation programs beidentified? Might such identification improve effortsto retain participants in the programs?hat design characteristics of home visitation pro-grams improve the work satisfaction and retention of
home visitors? l
hat background characteristics of visitors and re-quired pre-program training minimize visitor drop-out and maximize program performance?hat features of service systems are essential for effi-cient implementation and sustainability of homevisitation programs?hat is the minimum level of services infrastructureneeded to support adequate supervision of lay homevisitors?hat combination of community characteristics pro-vides optimal community readiness for implementa-tion and sustainability of home visitation programs?

iscussion

his review addresses the effects of early childhoodome visitation on child maltreatment and other vio-

ent outcomes. Substantial positive effects have beenound for the prevention of child maltreatment—a

edian relative reduction of 39%. This effect estimates most likely an underestimate, given that ascertain-

ent of violence in the intervention group may actuallye increased by the presence of the visitors. The

ntervention may also change long-term violent behav-ors by visited children and their parents, but thevidence related to those outcomes is not yet sufficiento draw conclusions or make recommendations. Manyther possible benefits may result from early homeisitations (as discussed above), and they all should bessessed when determining the ultimate cost-benefitalance of such interventions.The impressive beneficial effect despite the hetero-

eneity of home visitation programs in the Unitedtates—which often differ in their focus, curricula,uration, visitor qualifications, and target popula-ions—suggests the robustness of the home visitationntervention. It also raises the question of whetherhere is one optimal, effective, and cost-effective ap-roach for the multiplicity of possible outcomes. Thereater improvements found in our review for pro-rams using professional visitors (vs paraprofessionals)nd for programs of longer duration are only a start innswering a long list of research questions related tonding the best approaches for early childhood homeisitation.

Our findings and recommendations are similar tohose of some government and not-for-profit agencies,ut differ from findings and recommendations of oth-rs. The Canadian Task Force on Preventive Healthare32 and the Center for the Study and Prevention ofiolence (CSPV)27 recommend home visitation for therevention of child abuse in disadvantaged families, asoes the Community Guide. Other agencies, includinghe CDC’s National Center for Injury Prevention andontrol29; U.S. Surgeon General’s report, Youth Vio-

ence26; the report prepared for the U.S. Department of

Am J Prev Med 2005;28(2S1) 23

JWCtnrtmfirlfiwihaoc4

rdsgeqdvplor

nmpphmd

MRSADCfTbSNLJ

sCI

DtmFDBCISYH

i

PdCD

o

R

1

11

1

1

1

1

1

2

ustice’s Office of Justice Programs, Preventing Crime:hat Works, What Doesn’t, What’s Promising28; and theSPV all recommend home visitation for the preven-

ion of youth violence, delinquency, or crime. We haveot found sufficient evidence on which to base such aecommendation. Differences in recommendations arehe result of differing scope and methods of assess-

ent. The Surgeon General’s Report draws on thendings of CSPV, CDC, Preventing Crime, and othereviews. The CDC best practice recommendations areargely based on the assessments of specialists in theeld. The CSPV focuses on a single model program13

ith optimal evaluation design and replication. Prevent-ng Crime uses standards similar to ours, but assessesome visitation programs for children aged �2, as wells programs focused on the first 2 years, and considersutcomes that are not direct measures of violence (e.g.,ognitive development). In addition, our review coversmore years of research.Although home visitation is widespread among Eu-

opean nations (where these programs are usuallyelivered to all population groups), we found notudies evaluating violent outcomes in European pro-rams. Given that all of the evaluations we assessedxamined programs directed at high-risk populations, auestion remains: Should home visitation efforts beirected only to such populations, or might homeisitation be beneficial, and economically justified, foropulations at higher socioeconomic and educational

evels? The answer to this question may lie with benefitsf home visitation beyond the benefits of violenceeduction assessed in this review.

In conclusion, this review, along with the accompa-ying recommendations from the Task Force on Com-unity Preventive Services,74 should prove a useful and

owerful tool for public health policymakers, programlanners and implementers, and researchers. It canelp to secure resources and commitment for imple-enting home visitation interventions, and can provide

irection for further empirical research in this area.

embers of the systematic review development team wereobert A. Hahn, PhD, MPH, Oleg O. Bilukha, MD, PhD, andusan Snyder, PhD, Division of Prevention Research andnalytic Methods, Epidemiology Program Office, Centers forisease Control and Prevention (CDC), Atlanta GA; Alexrosby, MD, Division of Violence Prevention, National Center

or Injury Prevention and Control, CDC, Atlanta GA; Mindy. Fullilove, MD, New York State Psychiatric Institute, Colum-ia University, and the Task Force on Community Preventiveervices; Farris Tuma, ScD, and Eve K. Moscicki, ScD, MPH,ational Institute of Mental Health, Bethesda MD; and Akivaiberman, PhD, National Institute of Justice, Department of

ustice, Washington, DC.Members of the consultation team were Laurie M. Ander-

on, PhD, Epidemiology Program Office, CDC, Olympia WA;arl Bell, MD, Community Mental Health Council, Chicago

L; Red Crowley, Men Stopping Violence, Atlanta GA; Sujata


esai, PhD, National Center for Injury Prevention and Con-rol, CDC, Atlanta GA; Deborah French, Colorado Depart-

ent of Public Health and Environment, Denver CO; Darnell. Hawkins, PhD, JD, University of Illinois at Chicago;anielle LaRaque, MD, Harlem Hospital Center, New York;arbara Maciak, PhD, MPH, Epidemiology Program Office,DC, Detroit MI; James Mercy, PhD, National Center for

njury Prevention and Control, CDC, Atlanta GA; Suzannealzinger, PhD, New York State Psychiatric Institute, Nework; Patricia Smith, Michigan Department of Communityealth, Lansing.We are thankful for study abstractions conducted by Mel-

ssa Stigler, University of Minnesota, Minneapolis.Points of view are those of the Task Force on Community

reventive Services and of respective affiliated authors, ando not necessarily reflect those of the Centers for Diseaseontrol and Prevention; National Institute of Justice, U.S.epartment of Justice; or National Institutes of Health.No financial conflict of interest was reported by the authors

f this paper.

eferences1. Yoshikawa H. Long-term effects of early childhood programs on social

outcomes and delinquency. Future Child 1995;5:51–75.2. Barnett WS. Long-term effects of early childhood programs on cognitive

and school outcomes. Future Child 1995;5:25–50.3. Kamerman SB, Kahn A. Home health visiting in Europe. Future Child

1993;3:39–52.4. St.Pierre RG, Layzer JI, Barnes HV. Two-generation programs: design, cost,

and short-term effectiveness. Educ Child Dev 1995;5:76–93.5. Olds DL, Kitzman H, Cole K, Robinson JL. Theoretical foundations of a

program of home visitation for pregnant women and parents of youngchildren. J Community Psychol 1997;25:9–25.

6. Lally JR, Mangione PL, Honig AS. The Syracuse University Family Devel-opment Research Program: long-range impact of an early intervention withlow-income children and their families. In: Powell DR, ed. Parent educa-tion as early childhood intervention: emerging directions in theory,research and practice. Norwood, NJ: Alex Publishing Corporation,1988:79–104.

7. Bronfenbrenner U. The ecology of human development: experiments bynature and design. Cambridge MA: Harvard University Press, 1979.

8. Hawkins JD, Herrenkohl TI, Farrington DP, et al. Predictors of youthviolence. Washington DC: Office of Juvenile Justice and DelinquencyPrevention, U.S. Department of Justice, 2000 (Juvenile Justice Bulletin NCJ179065).

9. Bandura A. Self-efficacy: toward a unifying theory of behavioral change.Psychol Rev 1977;84:191–215.

0. Office of Juvenile Justice and Delinquency Prevention. Juvenile offendersand victims: 1999 national report. Washington DC: U.S. Department ofJustice, 1999.

1. Erikson E. Childhood and society. New York: W.W. Norton, 1950.2. Bowlby J. Attachment and loss. Vol. 1. Attachment. New York: Basic Books,

1969.3. Olds DL, Pettitt L, Robinson JL, et al. Reducing risks for antisocial behavior

with a program of prenatal and early childhood home visitation. J Com-munity Psychol 1998;26:65–83.

4. Centers for Disease Control and Prevention. Rates of homicide, suicide,and firearm-related death among children—26 industrialized countries.MMWR Morb Mortal Wkly Rep 1997;46:101–5.

5. Minino AM, Arias E, Kochanek KD, Murphy SL, Smith BL. Deaths: finaldata for 2000. Natl Vital Stat Rep 2002;50:35.

6. Maguire K, Pastore AL. Sourcebook of criminal justice statistics—1999.Washington DC: U.S. Department of Justice, Bureau of Justice Statistics,2000.

7. Council on Scientific Affairs, American Medical Association. Violenceagainst women. Relevance for medical practitioners. JAMA 1992;
267:3184–9.
ber 2S1

1

1

2

2

2

2

2

2

2

2

2

2

3

3

3

3

3

3

3

3

3

3

4

4

4

4

4

4

4

4

4

4

5

5

5

5

5

5

5

5

5

5

6

6

6

6

6

6

6

6

6

6

7

8. National Research Council. Understanding child abuse and neglect. Wash-ington DC: National Academy Press, 1993.

9. Sedlak AJ, Broadhurst DD. Third national incidence study of child abuseand neglect. Washington DC: U.S. Department of Health and HumanServices, Administration for Children and Families, National Center onChild Abuse and Neglect, 1996.

0. Dube SR, Anda RF, Felitti VJ, Chapman DP, Williamson DF, Giles WH.Childhood abuse, household dysfunction, and the risk of attempted suicidethroughout the life span: findings from the Adverse Childhood Experi-ences study. JAMA 2001;286:3089–96.

1. Kelley BT, Thornberry TP, Smith CA. In the wake of child maltreatment.Washington DC: Office of Juvenile Justice and Delinquency Prevention,U.S. Department of Justice, 1997.

2. Dodge KA, Bates JE, Pettit GS. Mechanisms in the cycle of violence. Science1990;250:1678–83.

3. Task Force on Community Preventive Services. Introducing the Guide toCommunity Preventive Services: methods, first recommendations, andexpert commentary. Am J Prev Med 2000;18(suppl 1).

4. U.S. Department of Health and Human Services. Healthy people 2010.Washington DC: U.S. Department of Health and Human Services, 2001.

5. Krugman RD. Universal home visiting: a recommendation from the U.S.Advisory Board on Child Abuse and Neglect. Future Child 1993;3:184–200.

6. U.S. Department of Health and Human Services. Youth violence: a reportof the Surgeon General. Washington DC: Department of Health andHuman Services, 2001.

7. Olds DL, Hill P, Mihalic SF, O’Brien R. Prenatal and infancy homevisitation by nurses. In: Elliott DS, ed. Blueprints for violence prevention.Boulder CO: Center for the Study and prevention of Violence, Institute ofBehavioral Science, University of Colorado at Boulder, 1998:1–97.

8. Sherman LW, Gottfredson DC, MacKenzie DL, Eck J, Reuter P, BushwaySD. Preventing crime: what works, what doesn’t, what’s promising. A reportto the United States Congress (NCJ 171676). Washington DC: U.S.Department of Justice, Office of Justice Programs, 1997.

9. Thornton TN, Craft CA, Dahlberg LL, lynch BS, Baer K. Best practices ofyouth violence prevention: a sourcebook for community action. AtlantaGA: Centers for Disease Control and Prevention, 2000.

0. Developmental Research and Programs, Inc. Communities that Careprevention strategies: a research guide to what works. Seattle WA: Devel-opmental Research and Programs, Inc., 2000.

1. Olds DL, Henderson CR Jr, Cole R, et al. Long-term effects of nurse homevisitation on children’s criminal and antisocial behavior: 15-year follow-upof a randomized controlled trial. JAMA 1998;280:1238–44.

2. MacMillan HL, Feightner JW, Goldbloom R, et al. Preventive health care,2000 update: prevention of child maltreatment. CAMJ 2000;163:1451–8.

3. Briss PA, Zaza S, Pappaioanou M, et al. Developing an evidence-basedGuide to Community Preventive Services—methods. Am J Prev Med2000;18(suppl 1):35–43.

4. Truman BI, Smith-Akin CK, Hinman AR, et al. Developing the Guide toCommunity Preventive Services—overview and rationale. Am J Prev Med2000;18(suppl 1):18–26.

5. Carande-Kulis VG, Maciosek MV, Briss PA, et al. Methods for systematicreviews of economic evaluations for the Guide to Community PreventiveServices. Am J Prev Med 2000;18(suppl 1):75–91.

6. Community Guide Economic Review Team. Economic evaluation abstrac-tion form (version 3.0). Available at: www.thecommunityguide.org/methods/default.htm. Accessed August 7, 2002.

7. Zaza S, Wright de Aguero L, Briss PA, et al. Data collection instrument andprocedure for systematic reviews in the Guide to Community PreventiveServices. Am J Prev Med 2000;18(suppl 1):44–74.

8. American Psychiatric Association. Diagnostic and statistical manual ofmental disorders (DSM-IV). Washington DC: American Psychiatric Publish-ing, Inc., 2002.

9. Achenbach TM, Howell CT, Aoki MF, Rauh VA. Nine-year outcome of theVermont intervention program for low birth weight infants. Pediatrics1993;91:45–55.

0. St. Pierre RG, Layzer JI. Using home visits for multiple purposes: theComprehensive Child Development Program. Future Child 1999;9:134–51.

1. Achenbach TM, Edelbrock C. Manual for the Child Behavior Checklist andRevised Child Behavior Profile. Burlington: University of Vermont, Depart-ment of Psychiatry, 1983.

2. Olds DL, Eckenrode J, Henderson CR Jr, et al. Long-term effects of homevisitation on maternal life course and child abuse and neglect: fifteen-year
follow-up of a randomized trial. JAMA 1997;278:637–43.
3. Eckenrode J. What works in nurse home visiting programs. In: AlexanderG, Curtis PA, Kluger MP, eds. What works in child welfare. Washington DC:Child Welfare League of America, Inc., 2000:35–43.

4. Ewigman B, Kivlahan C, Land G. The Missouri child fatality study:underreporting of maltreatment fatalities among children younger thanfive years of age, 1983 through 1986. Pediatrics 1993;91:330–7.

5. Armstrong KA. A treatment and education program for parents andchildren who are at-risk of abuse and neglect. Child Abuse Neglect1981;5:167–75.

6. Barth RP. An experimental evaluation of in-home child abuse preventionservices. Child Abuse Neglect 1991;15:363–75.

7. Brayden R, Altemeier W, Dietrich M, et al. A prospective study of secondaryprevention of child maltreatment. J Pediatr 1993;122:511–6.

8. Brooten D, Kumar S, Brown LP, et al. A randomized clinical trial of earlyhospital discharge and home follow-up of very-low-birth-weight infants.N Engl J Med 1986;315:934–9.

9. Caruso Whitney GA. Early intervention for high-risk families: reflecting ona 20-year-old model. In: Albee GW, Gullotta TP, eds. Primary preventionworks. Thousand Oaks CA: Sage, 1997:68–86.

0. Dawson P, Van Doornick WJ, Robinson JL. Effects of home-based, informalsocial support on child health. J Dev Behav Pediatr 1989;10:63–7.

1. Duggan A, Windham A, McFarlane E, et al. Hawaii’s healthy start programof home visiting for at-risk families: evaluation of family identification,family engagement, and service delivery. Pediatrics 2000;105:250–9.

2. Flynn L. The adolescent parenting program: improving outcomes throughmentorship. Public Health Nurs 1999;16:182–9.

3. Gray JD, Cutler CA, Dean JG, Kempe CH. Prediction and prevention ofchild abuse and neglect. J Social Issues 1979;35:127–39.

4. Hardy JB, Street R. Family support and parenting education in the home:an effective extension of clinic-based preventive health care services forpoor children. J Pediatr 1989;115:927–31.

5. Honig AS, Morin C. When should programs for teen parents and babiesbegin? Longitudinal evaluation of a teen parents and babies program. JPrimary Prev 2001;21:447–54.

6. Huxley P, Warner R. Primary prevention of parenting dysfunction in highrisk cases. Am J Orthopsychiatry 1993;63:582–8.

7. Katzev A, Pratt C, Henderson T, McGuigan W. Oregon’s Healthy Starteffort: 1997–98 status report. Corvallis: Oregon State University FamilyPolicy Program, 1999.

8. Kitzman H, Olds DL, Henderson JrCR , et al. Effect of prenatal and infancyhome visitation by nurses on pregnancy outcomes, childhood injuries, andrepeated childbearing: a randomized controlled trial. JAMA 1997;278:644–52.

9. Larson CP. Efficacy of prenatal and postnatal home visits on child healthand development. Pediatrics 1980;66:191–7.

0. Marcenko MO, Spence M, Samost L. Outcomes of a home visitation trialfor pregnant and postpartum women at-risk for child placement. ChildYouth Services Rev 1996;18:243–59.

1. Mulsow MH, Murry VM. Parenting on edge: economically stressed, single,African American adolescent mothers. J Fam Issues 1996;17:704–21.

2. Siegel E, Bauman KE, Schaefer ES, Saunders MM, Ingram DD. Hospitaland home support during infancy: impact on maternal attachment, childabuse and neglect, and health care utilization. Pediatrics 1980;66:183–90.

3. Velasquez J, Christensen M, Schommer B. Intensive services help preventchild abuse. Am J Maternal Child Nurs 1984;9:113–7.

4. Wagner MM, Clayton SL. The Parents as Teachers program: results fromtwo demonstrations. Future Child 1999;9:91–115.

5. Olds DL, Henderson Jr, CR Phelps C, Kitzman H, Hanks C. Effect ofprenatal and infancy nurse home visitation on government spending. MedCare 1993;31:155–74.

6. Eckenrode J, Ganzel B, Henderson CR Jr, et al. Preventing child abuse andneglect with a program of nurse home visitation: the limiting effects ofdomestic violence. JAMA 2000;284:1385–91.

7. Plotnick RD. Using benefit-cost analysis to assess child abuse preventionand intervention programs. Child Welfare 1999;78:381–403.

8. Duggan AK, McFarlane EC, Windham AM, et al. Evaluation of Hawaii’sHealthy Start program. Future Child 1999;9:66–90.

9. Barth R, Ash JR, Hacking S. Identifying, screening and engaging high-riskclients in private nonprofit child abuse prevention programs. Child AbuseNeglect 1986;10:99–109.

0. Spoth R, Redmond C, Hockaday C, Shin CY. Family programs: barriers toparticipation in family skills preventive interventions and their evaluation.
A replication and extension. Fam Relations 1996;43:247–54.
Am J Prev Med 2005;28(2S1) 25

www.thecommunityguide.org/PDFS/ECONEVAL_version3.pdf

www.thecommunityguide.org/PDFS/ECONEVAL_version3.pdf

7

7

7

7

2

1. Olds DL, Henderson CRJ. The prevention of maltreatment. In: Cicchetti D,Carlson V, eds. Child maltreatment: theory and research on the causes andconsequences of child abuse and neglect. New York: Cambridge UniversityPress, 1989:722–63.

2. Eberhardt MS, Ingram DD, Makuc DM, et al. Health, United States, 2001,with urban and rural chartbook. Hyattsville MD: National Center forHealth Statistics, 2001.


3. Olds DL, Henderson CRJ, Kitzman H, Eckenrode J, Cole R, Tatelbaum R.Prenatal and infancy home visitation by nurses: recent findings. FutureChild 1999;9:44–65.

4. Task Force on Community Preventive Services. Recommendations toreduce violence through early childhood home visitation, therapeuticfoster care, and firearms laws. Am J Prev Med 2005;28(suppl 1):1–97.

ber 2S1

Appendix ATable A1. Studies measuring effect of home visitation on preventing violence

Author & yearDesign suitability: designQuality of execution

LocationStudy periodPopulation

Intervention Sample selectionAssignment totreatment conditionsSample size (atassessment)

Effect measureEffect reported instudy

Value used insummary, relativepercent change(significance level)

Frequency and durationVisitor type

Other components(study arms, if any)

Violence by childAchenbach (1993)1

Highest: prospectivewith comparison

Good

Burlington VTRecruitment in

1980–1981Mean age of

mothers: 25–29 yr; ethnicitynot reported;only low-birthweight infants(birth weight:�2250g)

Four visits: at 3, 14, 30,and 90 days afterdischarge; initiatedpost-birth

Neonatal nurse

Seven 1-hoursessions byneonatal nurse inmaternityhospital, beforedischarge

All eligibleRandomized into

low-birth weightintervention andlow-birth weightcontrol groups

I: n�24C: n�31

Externalizing scale ofChild BehaviorChecklist (at age 9years)

No significantdifference betweenlow-birth weightintervention andlow-birth weightcontrol groups

No significantdifference

Lally (1988)2


Fair

Syracuse NYRecruitment in

1969–1971“Majority black”

families; meanage ofmothers: 18 yr,income�$5000 yr;85%unmarried

Visits: one/wk; initiatedin the 3rd trimesterof pregnancy,continued untilchild’s 5th birthday

Paraprofessional

Free child care 50wk/yr, for 5 yr(includingtransportation)

Parent supportgroups

Probablyconvenience (notdescribed)

Matched controlsrecruited whenintervention groupchildren were age3 yr

I: n�65C: n�54

Number of subjectsprocessed asprobation cases bythe countyprobationdepartment,assessed whenchildren were 13 to16 years of age(proportion ofsample)

Intervention group: 4(6%)

Control group: 12(22%)

(p�0.01)

�72.3% (p�0.01)

(continued on next page)

Am

JPrev

Med

2005;28(2S1)27

Table A1. (continued)








Olds (1998)3


Good

Elmira NYRecruitment

1978–198088% white; 11%

black; 62%unmarried;59% low SES;48% aged�19 yr

Visits: one/wk, thenless frequently,initiated before 29thwk of gestation,through child’s 2ndbirthday

Nurses

Both interventionand controlgroups receivedfreetransportation toprenatal andwell-child careanddevelopmentalscreening

All eligibleRandomizedI: n�97C: n�148

All outcomes assessedwhen childrenwere aged 15 yr

Major delinquentacts per subject(self-report):

Intervention group:3.57

Control group: 3.02(NS)

Incidence of arrests(self-report):


Control group: 0.36(p�0.01)

Incidence ofconvictions andprobationviolations(self-report):



Incidence of arrests(mother report):



Major delinquent acts:�18.2% (NS)

Arrests (self-report):�52.8% (p�0.01)

Convictions(self-report):�63.0% (p�0.01)

Arrests (motherreport): �33.3%


28A

merican

Journal

ofPreven

tiveM

edicine,

Volum

e28,

Num

ber2S1









St. Pierre (1999)4


Fair

USA, 21 sitesProbably

1992–199535% of mothers

aged �20 yr,43% black,26% Hispanic,26% white,3% NativeAmerican, 1%Asian; 83%with income�$10,000/yr

Two to four visits/mo;5 yr

Paraprofessionals

Probablyconvenience (notdescribed)

Not randomI: n�1507C: n�1544

Externalizing scale ofChild BehaviorChecklist (at age 5yr)

No significantdifference betweenlow-birth weightintervention andlow-birth weightcontrol groups

No significantdifference

Child maltreatmentBarth (1991)5


Good

CaliforniaProbably

1984–198845% white, 31%

Hispanic, 17%black, 7%other, 40%receivedAFDC, 70%income�$10,000;mean age 23.5yr; screened as“at risk” forchild abuse

About one every 2 wk,for 6 mo; mean 11visits (5–20), initiatedpost-birth

Paraprofessionals

Convenience(referrals)

RandomizedI: n�97C: n�94

Substantiated childabuse reports,before and afterthe program (atprogramcompletion)

Intervention group: 5(before) and 15(after)

Control group 1(before) and 14(after)

�23.1% (NS)

Brayden (1993)6

Highest prospective withcomparison

Fair

TennesseeProbably

1984–1989Mean age �22

yr; 70% white;all below200% ofpoverty level;screened as“at risk” forchild abuse

Frequency notreported; 0–2 yr,initiated post-birth

Probablyparaprofessionals(not clearly stated)

Pediatric carePsychological

supportcounseling


Substantiatedphysical abusereports asproportion of thesample (atprogramcompletion)

Intervention group:9.2%

Control group: 6.6%

�39.5% (NS)


Am

JPrev

Med

2005;28(2S1)29









Brooten (1986)7


Good

Pennsylvania1982–1986

�80% black,20% white;�2/3 income�$10,000;only very-low-birthweightinfants

Five visits from birth to18 mo, initiatedpost-birth

Nurses

On-call nurseavailability


Reported cases ofchild abuse (at 18mo)

Intervention group: 2Control group: 4

�48.7% (NS)

Caruso Whitney (1997)8


Fair

FloridaProbably

1977–1986Over 80% black,

6%–7% white;7%–12%Hispanic; lowSES; 75%unmarried

First yr: 1/wk, 2nd yr: 1every 2 wks, 3rd yr:usually 1/mo;initiated in the 3rdtrimester ofpregnancy, continueduntil age 3 yr

Mental health workers

Parent supportgroups, childactivity groups

Presumably alleligible

NonrandomI: n�171C: n�91

Reports of abuse orneglect resulting inremoval fromhome (at 12–18mo), proportion ofthe sample


Control group:7.79%

�93.2% (NS)

Dawson (1989)9


Fair

Colorado1977–1979

74% White, 25%Hispanic; lowSES

Median number ofcompleted homevisits: 30, initiated at30th wk ofpregnancy, continueduntil age 14 mo

Paraprofessionals(communitymothers)

Pediatric careParent support

groups once in 2weeks (only forhalf ofinterventiongroup)

ConvenienceRandomizedI: n�67C: n�44

Reports of potentialchild abuse orneglect


�228.4% (NS)


30A

merican

Journal

ofPreven

tiveM

edicine,

Volum

e28,

Num

ber2S1









Duggan (1999)10


Fair

Hawaii1994–1998

20% nativeHawaiian,13% PacificIslander, 19%Filipino, 12%Caucasian,7%–10%Asian, restmultiracial orunknown;2/3: belowpoverty level;screened as“at risk” forchild abuse

Starts at one/wk, thenless frequently;(mean number ofvisits: 13) from birthto age 2 years

Paraprofessionals


Confirmed childabuse and neglectreports to HawaiiChild ProtectiveServices (at 2 yr),proportion of thesample

Intervention group:2%

Control group: 3%

�33.3% (NS)

Flynn (1999)11

Lowest: post-interventionmeasure with nocomparison(compared withnational rate only)

Fair

New Jersey1995–1997

All mothersaged �19 yr;mean: 16.9 yr;71% black,27% Hispanic;screened as“at risk” forchild abuse

Starts at one/wk, thenone every 2 wks,one/mo, one every 3mo; duration notspecified, apparentlyabout 2 yrs, initiationat birth or prenatally

Paraprofessionals

ConvenienceNo controlI: n�137

Confirmed cases ofchild abuse andneglect

Intervention group:four cases (rate2.9%)

(Compares withnational rate byadolescent mothersof 11%)

�73.6% (NS)

Gray (1979)12


Fair

ColoradoProbably

1971–1975No information

on race,economicstatus, or age;screened as“at risk” forchild abuse

Visits one/wk; initiatedafter birth, durationunclear (evaluated at17 mo)

Nurses

Pediatric care Random samplefrom all eligible


Injury suspicious forabuse/neglect (at17 mo)


�100% (p�0.01)


Am

JPrev

Med

2005;28(2S1)31









Hardy (1989)13


Fair

Baltimore MDProbably

1983–1987All black, urban;

78% ofmothers single

Visits every 2 to 3 mo;from birth until 2 yr,about 8 to 10 total

Paraprofessionals


Randomized(odd-even)

I: n�131C: n�132

Suspected orconfirmed cases ofchild abuse andneglect(proportion ofsample), follow-up�10 mo (average22 to 24 mo)

Intervention group: 2(1.5%)

Control group: 13(9.8%)

�84.7% (NS)

Honig (2001)14


Fair

Syracuse NYStudy period not

reportedMean age 17.5

yr (range 13to 21); 95%receivingpublicassistance;race/ethnicitynot reported;screened as“at risk” forchild abuse

Visits one/wk; duration18 to 27 mo; initiatedprenatally (Arm A)or postnatally (ArmB)

Presumablyparaprofessionals

ConvenienceNot random

(control groupsformed fromprogramdropouts)

Arm A:I: n�52C: n�13Arm B:I: n�25C: n�24

Child abuse reports,proportion ofsample (apparentlyat exit, 18 to 27mo)

Arm A (pre-birth):Intervention group:

13.5%Control group:

15.4%Arm B (post-birth):Intervention group:

20%Control group:

54.2%

Arm A: �12.3% (NS)Arm B: �63.1%

(p�0.05)

Huxley (1993)15


Fair

Boulder COProbably

1987–1990Mean age: 21

(I), 19 (C);ethnicity notreported;screened as“at risk” forchild abuse

Frequency andduration notreported, evaluationat 12 mo

Paraprofessionals, alsopublic health nursesand mental healthprofessionals

ConvenienceAll eligible controls

included,matched to cases

I: n�20C: n�20

Confirmed reports ofchild abuse (at 12mo)


�75.0% (p�0.07)


32A

merican

Journal

ofPreven

tiveM

edicine,

Volum

e28,

Num

ber2S1









Katzev (1999)16


Fair

Oregon1996–1997

Mean age: 20.7yr, 29% aged�17 yr; 74%white, 22%Hispanicscreened as“at risk” forchild abuse

Visits one/wk, then lessfrequently,presumably frombirth to 2 yr

Paraprofessionals

Parent supportgroups

ConvenienceComparison with

those who refusedto participate

I: n�1332C: n�1372

Child abuse reports,proportion ofsample


Control group: 2.6%

�30.8% (NS)

Kitzman (1997)17


Good

Memphis TN1990–1993

92% black; 85%below poverty;mean age: 18yr

Visits one/wk, then lessfrequently; initiatedprenatally, 29 wkgestation to 2 yr;mean: 7 prenatal and26 postnatal visits

Nurses

Both interventionand controlgroups receivedfreetransportation topediatricappointmentsanddevelopmentalscreening


Incidence ofemergencydepartment visitsfor injury oringestion (throughage 2 yr)


Control group: 0.34

�2.9% (NS)

Larson (1980)18


Fair

Montreal,Canada

Study period notreported

White; “workingclass income”age: 18 to 35yr

Visits one every 2 wk,then less frequently;duration 15 mo,initiated prenatally(Arm A) or 6 wkpostnatally (Arm B),both groups receivedten visits total

Child psychologists

ConvenienceRandomized (both

Arm A and Arm Bcompared withthe same controlgroup)

I: (Arm A) n�26I: (Arm B) n�27C: n�37

Accident rate perchild (at 18 mo)

Arm A (pre-birth):Intervention group:

0.86Arm B (post-birth):Intervention group:

1.26Control group: 1.55

Arm A: �44.5%(p�0.01)

Arm B: �18.7% (NS)

Marcenko (1996)19


Fair

PennsylvaniaStudy period not

reported93%–95% black;

mean age �24yr; meanincome $450to $480 permo; screenedas “at risk” forout-of-homeplacement ofa child

Visits one every 2 wk;initiated in 3rd to 6thmo of pregnancy,ended when childaged 1 yr

Paraprofessionals



Out-of homeplacement (eitherin foster care orwithrelatives/friends)(measured at 16mo), proportion ofsample

Intervention group:26%

Control group: 23%

�13.0% (NS)


Am

JPrev

Med

2005;28(2S1)33









Mulsow (1996)20


Fair

Georgia1993–1995

All black,teenage (13 to20 yr), urban;89% receivedsome form ofgovernmentsupport;screened as“at risk” forstressfulconditions ofparenting

Duration and frequencyof visits not reported;initiated after birth

Presumablyparaprofessionals


Not randomI: n�15C: n�36

Reports of childabuse and neglect(both substantiatedandnonsubstantiated)(at 2 yr)



�94.1% (p�0.18)

Olds (1997)21


Good

Elmira NY1978–1982

88% white, 11%black; 62%unmarried;59% low SES;48% aged�19 yr

Visits one/wk, then lessfrequently; initiatedbefore 29 wk ofgestation, throughchild’s 2nd birthday

Nurses



Substantiated reportsof child abuse andneglect, incidence(birth to 15 yr)(reports per 15-yrfollow-up period)


Control group: 3.47

�46.3% (p�0.01)


34A

merican

Journal

ofPreven

tiveM

edicine,

Volum

e28,

Num

ber2S1









Siegel (1980)22


Fair

Greensboro NC1976–1978

About 25%white, race/ethnicity ofremainder notreported; 65%single; meanage: 20–21 yr

Nine visits during first 3mo of life

Paraprofessionals

Arm A:Uncomplicateddelivery, earlymother–childcontact (at least45 minutesduring first 3hours afterdelivery), andhome visitation

Arm B:Uncomplicateddelivery andhome visitation(no earlymother–childcontact)

Arm C:Complicateddelivery (infantsplaced inobservationnursery afterbirth) and homevisitation


Randomized (ArmsA and Bcompared withuncomplicateddelivery controls;Arm C comparedwith complicateddelivery controls)

Uncomplicateddelivery:

I (Arm A): n�47I (Arm B): n�53C: n�52Complicated

delivery:I (Arm C): n�60C: n�59

Reports of abuse andneglect, throughchild’s 1st birthday

Uncomplicateddelivery:

Arm A: 4Arm B: 7Control group: 3Complicated delivery:Arm C: 3Control group: 3

Arm A: �47.5% (NS)Arm B: �128.9%

(NS)Arm C: �1.7% (NS)

Velasquez (1984)23


Fair

MinnesotaStudy period not

reportedAge: 16 to 25

years; ethnicitynot reported;majorityunmarried;screened as“at risk” forchild abuse

Visits one/wk; durationat least 18 mo,initiated after birth

Nurses, social workers

Intensive healthcare services

Convenience sampleNot randomI: n�23C: n�32

Number of infantswho experiencedone or moreoccurrences ofabuse, neglect, orout-of-homeplacement,through 18 mo(proportion ofsample)



�78.0% (p�0.01)


Am

JPrev

Med

2005;28(2S1)35









Wagner (1999)24


Fair

CaliforniaProbably

1991–1994All teenagers

aged �19 yrabout 1/3receivingAFDC; �50%Hispanic

Visits one/mo; duration2 yr, initiated beforebirth or after birth inthe first 6 mo of life

Paraprofessionals

Arm A: Homevisitation andparent supportgroup

Arm B: Homevisitation, casemanagement,and parentsupport group

ConvenienceRandomized (both

arms comparedwith the samecontrols [notreatment])

I (Arm A): n�149I (Arm B): n�138C: n�163

Cases of abuse andneglect (at 2 yr),proportion ofsample

Arm A: 1.3%Arm B: 0%Control: 2.4%

Arm A: �45% (NS)Arm B: �100%

(p�0.05)

Wagner (1999)24


Fair

California1992–1996

�75% Hispanic:about 20%receivingAFDC

Visits one/mo; duration3 yr, initiated afterbirth in first 6 mo oflife

Paraprofessionals

Parent supportgroup available


Treated for injuryduring past year(assessed at age 3),proportion ofsample


Control group:11.9%

�31.9% (NS)


36A

merican

Journal

ofPreven

tiveM

edicine,

Volum

e28,

Num

ber2S1









Violence by parentsOlds (1997)21


Good

Elmira NYRecruitment

1978–198088% white; 11%

black; 62%unmarried;59% low SES;48% aged�19 yr

Visits one/wk, then lessfrequently, initiatedbefore 29th wk ofgestation throughchild’s 2nd birthday

Nurses


All eligibleRandomizedTotal sample:I: n�97C: n�148Subsample including

only unmarried,low-SES mothers:

I: n�38C: n�62

(All outcomes relateto children’smothers, assessed15 yr afterinterventionstarted)

Incidence of arrests(state records)

Total sample:Intervention group:


(NS)Low SES unmarried:Intervention group:


(p�0.01)Incidence of arrests

(self-report)Total sample:Intervention group:




(p�0.01)Incidence of

convictions (staterecords)



(NS)

Arrests (state records)Total sample: �68.4%

(NS)Low SES, unmarried:

�82.2% (p � 0.01)Arrests (self-report)Total sample: �59.1%


�69.0% (p�0.01)Convictions (state

records)Total sample: �55.6%

(NS)Low SES unmarried:

�81.8% (p � 0.01)Convictions

(self-report)Total sample: �76.9%


�78.4% (p � 0.01)


Am

JPrev

Med

2005;28(2S1)37









Low SES unmarried:Intervention group:


(p�0.01)Incidence of

convictions(self-report):





(p�0.01)Intimate partner

violenceEckenrode (2000)25


Good

Elmira, NewYork

Recruitment1978–1980

88% white, 11%black; 62%unmarried;59% low SES;48% aged�19 years

Visits 1/wk, then lessfrequently, initiatedbefore 29th wk ofgestation, throughchild’s 2nd birthday

Nurses


All eligibleRandomizedTotal sample:I: n�97C: n�148

Incidence ofdomestic violenceover the 15-yearfollow-up period

No significantdifference betweenintervention andcontrol groups

No significantdifference, no valuespecified

AFDC, Aid to Families with Dependent Children; C, comparison; I, intervention; mo, month(s); n, sample size; NS, not significant; SES, socioeconomic status; wk, week(s); yr, year(s).

38A

merican

Journal

ofPreven

tiveM

edicine,

Volum

e28,

Num

ber2S1

R
eferences for the Appendix
1. Achenbach TM, Howell CT, Aoki MF, Rauh VA. Nine-year outcome of the Vermont intervention programfor low birth weight infants. Pediatrics 1993;91:45–55.

2. Lally JR, Mangione PL, Honig AS. The Syracuse Uni-versity Family Development Research Program: long-range impact of an early intervention with low-incomechildren and their families. In: Powell DR, Sigel IE,eds. Parent education as early childhood intervention:emerging directions in theory, research and practice.Norwood NJ: Alex Publishing Corporation, 1988:;–.

3. Olds DL, Henderson CR Jr, Cole R, et al. Long-termeffects of nurse home visitation on children’s criminaland antisocial behavior: 15-year follow-up of a random-ized controlled trial. JAMA 1998;280:1238–44.

4. St. Pierre RG, Layzer JI. Using home visits for multiplepurposes: Comprehensive Child Development Pro-gram. Future Child 1999;9:134–51.

5. Barth RP. An experimental evaluation of in-home childabuse prevention services. Child Abuse Neglect 1991;15:363–75.

6. Brayden R, Altemeier W, Dietrich M, et al. A prospec-tive study of secondary prevention of child maltreat-ment. J Pediatr 1993;122:511–6.

7. Brooten D, Kumar S, Brown LP, et al. A randomizedclinical trial of early hospital discharge and homefollow-up of very-low-birth-weight infants. N Engl J Med1986;315:934–9.

8. Caruso Whitney GA. Early intervention for high-riskfamilies: reflecting on a 20-year-old model. In: AlbeeGW, Gullotta TP, eds. Primary prevention works. Thou-sand Oaks, CA: Sage, 1997:68–86.

9. Dawson P, Van Doornick WJ, Robinson JL. Effects ofhome-based, informal social support on child health.Dev Behav Pediatr 1989;10:63–7.

10. Duggan AK, McFarlane EC, Windham AM, et al. Eval-uation of Hawaii’s Healthy Start program. Future Child1999;9:66–90.

11. Flynn L. The adolescent parenting program: improv-ing outcomes through mentorship. Public Health Nurs1999;16:182–9.

12. Gray JD, Cutler CA, Dean JG, Kempe CH. Predictionand prevention of child abuse and neglect. J SocialIssues 1979;35:127–39.

13. Hardy JB, Street R. Family support and parenting
education in the home: an effective extension of
clinic-based preventive health care services for poorchildren. J Pediatr 1989;115:927–31.

14. Honig AS, Morin C. When should programs for teenparents and babies begin? Longitudinal evaluation of ateen parents and babies program. J Primary Prev2001;21:447–54.

15. Huxley P, Warner R. Primary prevention of parentingdysfunction in high risk cases. Am J Orthopsychiatry1993;63:582–8.

16. Katzev A, Pratt C, Henderson T, McGuigan W. Ore-gon’s Healthy Start effort: 1997–98 status report. Cor-vallis, OR: Oregon State University Family Policy Pro-gram, 1999.

17. Kitzman H, Olds DL, Henderson CR Jr, et al. Effect ofprenatal and infancy home visitation by nurses onpregnancy outcomes, childhood injuries, and repeatedchildbearing: a randomized controlled trial. JAMA1997;278:644–52.

18. Larson CP. Efficacy of prenatal and postnatal homevisits on child health and development. Pediatrics1980;66:191–7.

19. Marcenko MO, Spence M, Samost L. Outcomes of ahome visitation trial for pregnant and postpartumwomen at-risk for child placement. Child Youth Ser-vices Rev 1996;18:243–59.

20. Mulsow MH, Murry VM. Parenting on edge: economi-cally stressed, single, African American adolescentmothers. J Fam Issues 1996;17:704–21.

21. Olds DL, Eckenrode J, Henderson CR Jr, et al. Long-term effects of home visitation on maternal life courseand child abuse and neglect: fifteen-year follow-up of arandomized trial. JAMA 1997;278:637–43.

22. Siegel E, Bauman KE, Schaefer ES, Saunders MM,Ingram DD. Hospital and home support during in-fancy: impact on maternal attachment, child abuse andneglect, and health care utilization. Pediatrics 1980;66:183–90.

23. Velasquez J, Christensen M, Schommer B. Intensiveservices help prevent child abuse. Am J Maternal ChildNurs 1984;9:113–7.

24. Wagner MM, Clayton SL. The Parents as Teachersprogram: results from two demonstrations. FutureChild 1999;9:91–115.

25. Eckenrode J, Ganzel B, Henderson CR Jr, et al. Pre-venting child abuse and neglect with a program ofnurse home visitation: the limiting effects of domestic
violence. JAMA 2000;284:1385–91.
Am J Prev Med 2005;28(2S1) 39

The Effectiveness of Early Childhood Home … Effectiveness of Early Childhood Home Visitation in...

Documents

Transcript of The Effectiveness of Early Childhood Home … Effectiveness of Early Childhood Home Visitation in...