02 Qian Lu (bc/d)
21/1/05
3:40 pm
Page 185
Learning from
Success and Failure
in Psychosocial
Intervention:
An Evaluation of
Low Birth Weight
Prevention Trials
QIAN LU, MICHAEL C. LU, &
CHRISTINE DUNKEL SCHETTER
University of California, Los Angeles, USA
is a doctoral student in training in Social
and Health Psychology at UCLA. She received her
MD degree from Shandong Medical University. Her
research focuses on the evaluation of health
interventions, and health interventions for ethnic
minorities.
QIAN LU
M I C H A E L L U is an Assistant Professor in Obstetrics
and Gynecology and in the School of Public Health at
UCLA. Dr Lu received his MD degree from the
University of California, San Francisco, and MPH and
MS degrees from the University of California,
Berkeley. His current work is on ethnic disparities in
maternal and child outcomes.
C H R I S T I N E D U N K E L S C H E T T E R is a Professor of
Psychology and Director of Health Psychology at
UCLA where she has been on the faculty since 1983.
She received her PhD from Northwestern University
and was a postdoctoral fellow at the University of
California, Berkeley. Her research concerns stress
processes in pregnancy, social support, coping and
health outcomes.
COMPETING INTERESTS:
ADDRESS.
None declared.
Correspondence should be directed to:
P R O F. C H R I S D U N K E L S C H E T T E R , Department of Psychology,
University of California Los Angeles, 405 Hilgard Ave, Los Angeles,
CA 90095–1772, USA. [email: dunkel@psych.ucla.edu]
Journal of Health Psychology
Copyright © 2005 SAGE Publications
London, Thousand Oaks and New Delhi,
www.sagepublications.com
Vol 10(2) 185–195
DOI: 10.1177/1359105305049763
Abstract
The object of the study was to
evaluate the research designs of
social support interventions for
prevention of low birth weight
(LBW). A literature search of
published articles identified 12
randomized controlled trials of
social support to prevent LBW
birth. These were evaluated
using specific methodological
criteria for effective
intervention research. Only one
study showed a significant
reduction in LBW. However,
none of the studies met all of
the proposed criteria for
rigorous intervention research.
It is premature to conclude that
social support interventions are
ineffective in preventing LBW.
Specific recommendations for
future intervention research
design are outlined.
Keywords
intervention, low birth weight
(LBW), pregnancy risk,
psychosocial, social support
185
02 Qian Lu (bc/d)
21/1/05
3:40 pm
Page 186
JOURNAL OF HEALTH PSYCHOLOGY 10(2)
L OW B I RT H weight (LBW) is a leading cause of
infant mortality and morbidities in the USA
(Mathews, Menacker, & MacDorman, 2003).
Limited success in reducing LBW through
medical interventions (Lu, Tache, Alexander,
Kotelchuck, & Halfon, 2003), coupled with
accumulating evidence that links psychosocial
factors to pregnancy outcomes (Paarlberg,
Vingerhoets, Passchier, Dekker, & van Geijn,
1995) have led to interest in preventing LBW by
implementation of psychosocial interventions.
However, most programs that offer social
support during pregnancy have not been
particularly effective in preventing LBW
(Hodnett & Fredericks, 2003; Stevens-Simon &
Orleans, 1999). The primary aim of this article
is to evaluate the design of these intervention
studies. We contend that because most studies:
(1) have not been theory-driven; (2) have not
had effective risk screening; (3) have not
matched interventions to risks; and (4) have not
tested process variables, it may be premature to
conclude that psychosocial interventions are
ineffective in preventing LBW. In this article we
propose a new model of psychosocial intervention research for the prevention of LBW.
Methods
We conducted a literature search for original
research, systematic reviews, meta-analyses and
commentaries that were published in the English
language from 1960 to 2004 in computerized
databases, specifically, Pubmed, PsycInfo and
the Cochrane Library. Keywords used to
conduct the search included ‘low birth weight’,
‘preterm’, ‘randomized trials’, ‘prenatal care’,
‘social support’ ‘psychosocial intervention’
‘psychological intervention’ and/or ‘prevention’.
Additionally, the reference lists of all qualifying
articles and several recent reviews (e.g. Hodnett
Screen
population at
risk for LBW
& Fredericks, 2003; Stevens-Simon & Orleans,
1999) were examined for additional studies not
identified through the database search. Our
search generated 2279 articles. All abstracts
were reviewed, and all articles that could not be
conclusively excluded based on the abstract
were retrieved to determine eligibility. The
following criteria were used to select interventions to be included in this review: (1) the
intervention must be a randomized controlled
trial (RCT); (2) the intervention must be aimed
at preventing LBW; and (3) the intervention
must be mainly psychosocial in nature. We
excluded interventions that were primarily
medical (e.g. screening and treatment of
asymptomatic bacteriuria) or nutritional in
nature. We also exclude smoking cessation
interventions even if they included psychosocial
components. Based on these criteria, 12 randomized controlled trials were included in our review.
These 12 studies were evaluated using a new
model for LBW prevention intervention
research. The traditional model of psychosocial
intervention consists of three steps: screen the
population at risk for LBW; provide psychosocial support to at-risk populations; and then
compare LBW rates between intervention and
control groups (Fig. 1).
This traditional model has several pitfalls.
First, the intervention is often not theory-driven.
Without a predictive model linking intervention
to outcome, it is sometimes difficult to know
why, for whom and under what circumstances
the program works or does not work (Drotar,
2002). Second, risk screening is based on risk
factors for LBW, no matter whether or not they
are modifiable by psychosocial interventions.
Third, because some subjects are selected based
on medical risks rather than psychological risks
for LBW, psychosocial interventions may prove
inappropriate and ineffective. Lastly, the
Provide
psychosocial
support
Figure 1. Traditional model of psychosocial intervention for LBW prevention.
186
Assess
outcomes
(LBW)
02 Qian Lu (bc/d)
21/1/05
3:40 pm
Page 187
LU ET AL.: LEARNING FROM SUCCESS AND FAILURE
traditional model often fails to measure process
variables. The term process variables, as used
here, refers to variables mediating the effects of
interventions (Aiken, 1999), and is used interchangeably with the terms ‘mediators’ in this
article. Testing process variables allow the
investigators to better understand the processes
by which the intervention produces its effect, or
the reasons for the absence of any effects.
To address these pitfalls, we propose a new
model of psychosocial intervention research
for LBW prevention. Elaborating on a model
developed by Aiken (1999), and West and
Aiken (1997), our model consists of five stages.
The first stage is to construct a predictive model
that is empirically supported. The second stage
is to screen population based on risk factors
that are modifiable by psychosocial intervention. The third stage is to match interventions to risks. Psychosocial support should
be provided to women and families at increased
psychosocial risk, but not necessarily increased
medical risk, for LBW. The fourth stage is to
test process variables. Excellent examples of
testing process variables exist in the literature
for mammography screening (Aiken, West,
Woodward, Reno, & Reynolds, 1994) and other
issues such as AIDS prevention (Fisher, Fisher,
Misovich, Kimble, & Malloy, 1996) but this
approach has yet to be applied to LBW prevention. Statistical testing of mediators is beyond
the scope of this article and can be read in
Aiken et al. (1994) and Baron and Kenny
(1986). The final stage is to assess outcomes
(see Fig. 2). Thus the new model improves on
the traditional model by: (1) calling for a priori
development of a predictive model; (2) limiting
risk screening to modifiable psychosocial risk
factors; (3) matching psychosocial interventions
Develop a
predictive
model
Screen
population
based on
modifiable
psychosocial
risk factors
to psychosocial risks for LBW; and (4) testing
process variables.
Based on this new model, we established seven
criteria for evaluating the design of psychosocial
intervention research for LBW prevention:
1. Did the study describe a predictive model?
2. Did the study evaluate the effectiveness of
risk screening?
3. Was the risk screening effective in identifying women at risk for LBW?
4. Did the study target risk factors modifiable
by psychosocial interventions?
5. Did the study test process variables?
6. Did the intervention modify the process variables?
7. Did the study evaluate the relationship of
process variables to outcome (LBW)?
Results
Table 1 summarizes the characteristics of the
12 studies included in our review. All studies
offered social support as a single intervention or
the centerpiece of a multi-component prenatal
care intervention with a major goal of preventing LBW. Five studies also aimed to prevent
preterm delivery (PTD), one of the twin
contributors to LBW (the other being intrauterine growth restriction (IUGR) or fetuses
that are small-for-gestational-age (SGA),
defined as fetal or birth weight below the 10th
percentile for a given gestation) (Kitzman et al.,
1997; Klerman et al., 2000; Moore et al., 1998;
Olds, Henderson, Tatelbaum, & Chamberlin,
1986; Villar et al., 1992). Four studies had other
aims such as preventing child maltreatment
(McLaughlin et al., 1992) or improving developmental outcomes or pregnancy outcomes other
Match
interventions
to risks
Test process
variables
Assess
outcomes
(LBW)
Figure 2. New model of psychosocial intervention for LBW prevention.
187
C
A
L
L
L
B
W
P
T
D
O
T
H
M
E
D
P
S
Y
S
E
S
B
E
H
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
–
+
+
+
+
+
+
+
+
+
+
_
+
+
+
+
+
+
Content of support
Outcome (% LBW)
A
P
P
I
N
F
E
M
O
E
D
U
Intervention
group
(%)
Control
group
(%)
p value
+
+
+
+
+
+
+
+
+
+
12.5
10.9
15
9.1
8.7
10.5
10
14.0
19
19
11.2
14
14
22.4
9.4
7.5
9
11.6
22
20.5
NS
NS
NS
p = 0.005
NS
NS
NS
NS
NS
NS
+
8.8
5.8
8.6
2.6
+
+
+
H
O
M
E
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
I
N
S
+
+
+
+
NS
NS
Notes:
Goal of intervention: LBW = low birth weight (defined as weight <2500g), PTD = preterm delivery, OTH = others goals
Study population: W = White, AA = African American, LA = Latina American
Risk screening based on categories of risk factors: MED = medical, PSY = psychological, SES = socioeconomic status, BEH = behavioral;
‘–’ signifies risk as exclusion criteria, ‘+’ signifies risk as inclusion criteria
Format of social support: CALL = telephone call, HOME = home visit; APP = additional appointment
Content of social support: INF = informational support (information about available community services, signs of medical complications of pregnancy);
EMO = emotional/relational social support (provide emotional support directly or encourage emotional support from existing social network); INS = instrumental
social support (help directly with housing, shopping, other domestic work); EDU = health education
Outcome defined as percent LBW: NS = non-significant
Page 188
Spencer et al. (1989)
Olds et al. (1986)
Format of
support
3:40 pm
Ethnicity
AA
77% AA
92% AA
AA
LA
AA
AA & W
W
95% W
45% W &
53% AA
Non-Asian
W
Risk screening
21/1/05
Author
Klerman et al. (2000)
Moore et al. (1998)
Kitzman et al. (1997)
Norbeck et al. (1996)
Villar et al. (1992)
Graham et al. (1992)
McLaughlin et al. (1992)
Rothberg & Lits (1991)
Oakley et al. (1990)
Heins et al. (1990)
Goal
02 Qian Lu (bc/d)
Population
JOURNAL OF HEALTH PSYCHOLOGY 10(2)
188
Table 1. Characteristics of randomized controlled trials of psychosocial interventions for the prevention of LBW
02 Qian Lu (bc/d)
21/1/05
3:40 pm
Page 189
LU ET AL.: LEARNING FROM SUCCESS AND FAILURE
than LBW and PTD (Kitzman et al., 1997;
Klerman et al., 2000; Olds, Henderson, Tatelbaum, & Chamberlain, 1986).
Most of the studies recruited low-income
women or African-American women who were
at high risk of LBW. Among the twelve studies,
three prevention programs recruited solely
African-American women (Graham, Frank,
Zyzanski, Kitson, & Reeb, 1992; Klerman et al.,
2000; Norbeck, Dejoseph, & Smith, 1996), and
two studies recruited primarily African Americans (Kitzman et al., 1997; Moore et al., 1998).
Two studies had about half African Americans
and half Whites (Heins, Webster, McCarthy, &
Melvin, 1990; McLaughlin et al., 1992), and
three studies had primarily or solely Whites
(Oakley, Rajan, & Grant, 1990; Olds et al., 1986;
Rothberg & Lits, 1991). One study was
conducted with primarily Whites in Manchester,
England (Spencer, Thomas, & Morris, 1989),
and another study was conducted with Latina
Americans (Villar et al., 1992) (Table 1).
All of the 12 studies screened participants
based on risk factors, except the Olds et al.
(1986) study in which any women who asked to
participate were allowed to enroll in the study.
Risk factors in the studies can be classified into
four categories: medical, psychological, socioeconomic and behavioral (Table 1). The intervention groups in these studies were provided
with various formats of social support, such as
telephone calls (Moore et al., 1998), home visitations from nurses or social workers (Graham
et al., 1992; Kitzman et al., 1997; McLaughlin
et al., 1992; Olds et al., 1986; Spencer et al., 1989;
Villar et al., 1992), additional face-to-face
appointment in clinics (Heins et al., 1990;
Klerman et al., 2000; Rothberg & Lits, 1991) or
a combination of telephone calls and home visits
(Norbeck et al., 1996; Oakley et al., 1990). The
contents of social support provision included
informational support to help to recognize signs
and symptoms of pregnancy complications and
provide information about the utilization of
local services; emotional support to enhance the
existing social network and reduce stress
(Norbeck et al., 1996); instrumental support to
directly support domestic tasks (Spencer et al.,
1989); health education including nutrition
education and smoking cessation; and a combination of two or three of these intervention
components (Graham et al., 1992; Heins et al.,
1990; Klerman et al., 2000; McLaughlin et al.,
1992; Moore et al., 1998; Oakley et al., 1990;
Olds et al., 1986; Rothberg & Lits, 1991; Spencer
et al., 1989; Villar et al., 1992). The control
groups in these studies received standard or
usual care, except for two prevention programs
that offered free transportation to the control
group (Kitzman et al., 1997; Olds et al., 1986).
Only one (Norbeck et al., 1996) of the twelve
studies showed a significant reduction in LBW
rates (Table 1).
Table 2 summarizes our evaluation of the
intervention designs. Five of the twelve trials
(Graham et al., 1992; Heins et al., 1990; Klerman
et al., 2000; Moore et al., 1998. Norbeck et al.,
1996) developed a predictive model, or at least
had a simple predictive hypothesis (column 2).
Only two studies (Graham et al., 1992; Klerman
et al., 2000) evaluated the effectiveness of their
risk screening, none of which actually were effective (column 3). We evaluated the effectiveness
of risk screening for the remaining nine studies
by observing the LBW rates in the control
groups. If the risk screening was effective, then
the LBW rate in the control groups (with no
intervention) should be substantially higher than
that in the general population. That is, we would
expect LBW rates in the control groups of
greater than 7 percent in studies with predominantly White population and 14 percent in
studies with predominantly African-American
study population (Mathews, Menacker, &
MacDorman, 2003). For the two studies (Heins
et al., 1990; McLaughlin et al., 1992) with about
even population admixture of African Americans and Whites, the LBW rate in the control
group should be substantially higher than 10.5
percent if risk screening was effective. For the
study by Spencer et al. (1989), we would expect
the LBW rate in the control group of greater than
10.2 percent, the LBW rate in the general population in Manchester at the time of the study.
Similarly, for the study by Villar et al. (1992) with
Latina Americans, the LBW rate of 9.4 percent
in the control group was substantially lower than
the expected rate of 15 percent. As Table 2 illustrates, only four of the twelve studies (Heins et
al., 1990; Norbeck et al., 1996; Oakley et al., 1990;
Rothberg & Lits, 1991) met the criteria for effective risk screening (column 4). In other words,
the women studied in eight of twelve studies did
not appear to be at high risk for LBW.
189
Y
Y
N
Y
N
Y
N
N
N
Y
N
N
Y
N
N
N
N
Y
N
N
N
N
N
N
N
N
N
Y
N
N
N
Y
Y
Y
N
N
Y
N
N
Y
N
Y
N
Y
N
N
N
N
Notes: Y = Yes, N = No, NS = Non-significant, NK = Not known
8. Did the study
evaluate the
relationship of
process variables
to LBW?
Y
N
N
N
N
N
N
N
N
N
N
N
Y
NK
NK
NK
N
NK
NK
NK
NK
NK
NK
Y
N
NK
NK
NK
NK
NK
NK
NK
NK
NK
NK
N
Page 190
Klerman
Moore
Kitzman
Norbeck
Villar
Graham
McLaughlin
Rothberg
Oakley
Heins
Spencer
Olds
6. Did the study
test process
variables?
7. Did the
intervention
change process
variables?
3:40 pm
5. Did the study
target risk factors
modifiable by
psychosocial
interventions?
21/1/05
4. Was the risk
screening effective
in identifying
women at risk
for LBW?
02 Qian Lu (bc/d)
1. First author
2. Did the study
describe a
predictive model?
3. Did the study
evaluate the
effectiveness of
risk screening?
JOURNAL OF HEALTH PSYCHOLOGY 10(2)
190
Table 2. Evaluation of the study designs of psychosocial interventions for the prevention of LBW
02 Qian Lu (bc/d)
21/1/05
3:40 pm
Page 191
LU ET AL.: LEARNING FROM SUCCESS AND FAILURE
Of the twelve trials, only four (Graham et al.,
1992; Klerman et al., 2000; Norbeck et al., 1996;
Rothberg & Lits, 1991) screened for psychosocial risk factors for LBW that are modifiable
by psychosocial interventions; consequently
only these four are able to match psychosocial
interventions to psychosocial risks (column 5).
Only one (Klerman et al., 2000) tested process
variables (column 6). Two others (Olds et al.,
1986; Villar et al., 1992) assessed process variables in their study designs, but did not include
these process variables in the data analyses
(column 6). Of the three studies that have information about process variables, two (Klerman
et al., 2000; Olds et al., 1986) showed that the
intervention was effective in changing process
variables (column 7), though in neither study
were these variables evaluated in terms of their
relationship with outcome (LBW) (column 8).
Discussion
Of the 12 studies we reviewed, only one showed
a significant reduction in LBW rates. However,
before concluding that psychosocial interventions are ineffective in preventing LBW, it
is important to evaluate the designs of these
studies. Based on the work by Aiken (1999),
Drotar (2002), Stevens-Simon and Orleans
(1999) and West and Aiken (1997), we proposed
a new model for evaluating LBW-psychosocial
intervention research. None of the 12 studies
met all of the criteria for effective research
design. We found several pitfalls common to the
design of these studies.
Most interventions were
not theory-driven
The majority of the studies did not have a
predictive model (or even a simple predictive
hypothesis) that was grounded in social support
theory and supported by empirical data. A
predictive model specifies how the psychosocial
intervention is supposed to work. Without it, it
is sometimes difficult to know why, for whom
and under what circumstances the program
works or does not work (Drotar, 2002). Of the
seven studies that did not specify a predictive
model, only one (Rothberg & Lits, 1991)
conducted risk screening using psychosocial risk
factors and matched intervention to psychosocial risks, and none of these seven tested
psychosocial mediators (process variables) of
intervention effects. An effective intervention
design begins with a predictive model linking
intervention to outcome.
The sole successful intervention trial (Norbeck
et al., 1996) had a clear theoretical and empirical basis. It was hypothesized that in a population of pregnant African-American women
with inadequate social support, a culturally
competent social support intervention would
decrease the rate of LBW. The empirical basis
for the hypothesis came from an earlier study
(Norbeck & Anderson, 1989), which showed
that social support from the pregnant woman’s
mother and male partner, but not from female
friends or relatives, was directly predictive of
pregnancy outcomes. Thus, inadequate support
from mother and male partner was used as the
basis for identifying high-risk status. In order to
design an effective social support intervention,
the researchers used focus groups to learn
about the life situations of low-income AfricanAmerican pregnant women and the characteristics of the social support they needed during
pregnancy. They found that the essential
support from the woman’s mother or male
partner was emotional support. Based on focus
groups’ results, Norbeck et al. (1996) designed
an intervention to provide emotional support.
Thus, a predictive model that was grounded in
social support theory and supported by empirical evidence helped guide the design of an effective intervention program.
Most interventions did not have
effective risk screening
Risk screening is a critical first step in an
intervention trial. By identifying populations at
risk for LBW, interventions can be more effectively targeted to ‘high-risk’ populations. Our
review suggests that most intervention trials
have not had effective risk screening. Although
all twelve studies conducted risk assessment,
only two (Graham et al., 1992; Klerman et al.,
2000) evaluated the effectiveness of risk
screening, namely, whether the people who
were identified at risk were in fact at risk. In one
trial (Graham et al., 1992), for example, a
screening questionnaire containing medical and
psychosocial questions was used at the beginning of the study to differentiate between
participants hypothesized to be at high or low
191
02 Qian Lu (bc/d)
21/1/05
3:40 pm
Page 192
JOURNAL OF HEALTH PSYCHOLOGY 10(2)
risk of having a LBW infant. Risk factors
included low family function as assessed
by the Family APGAR Scale (Smilkstein, 1978),
the Modified Index of Family Relationships
(Hudson, 1982), stressful life events and
additional risk factors such as smoking, low
maternal weight–height ratio, age 27 or older
and history of a previous preterm delivery.
Results indicated that untreated subjects judged
by the questionnaire to be at high risk by the
screening instrument actually had a lower LBW
rate (9%) than those judged to be at low risk
(11%). Thus, the risk screening employed by
this study was not effective in identifying
women at risk for LBW births. In another study
(Olds et al., 1986) the LBW rate was so low in
the control group (2.6%) that it was nearly
impossible to show any effect of the intervention even though the social support manipulation seemed effective in that study.
Most studies screened for medical, socioeconomic and behavioral risk factors for LBW,
many of which were not modifiable by psychosocial interventions. Only four studies (Graham
et al., 1992; Klerman et al., 2000; Norbeck et al.,
1996; Rothberg & Lits, 1991) screened for
psychosocial risk factors; of these, only two
(Norbeck et al., 1996; Rothberg & Lits, 1991)
used screening instruments that were predictive
of LBW. One study (Graham et al., 1992) used
stressful life events as a screening instrument.
Several studies have shown that stressful life
events do not reliably predict LBW (Hoffman
& Hatch, 1996; Lobel, Dunkel-Schetter, &
Scrimshaw, 1992). In the only successful intervention trial, Norbeck et al. (1996) targeted
African-American women at risk for LBW due
to inadequate social support. Instead of comprehensive risk screening for medical, socioeconomic and behavioral risk factors, they used
the Norbeck Social Support Questionnaire
(NSSQ) to assess the adequacy of social
support. If the social support from either the
women’s mother or husband/partner was inadequate, women were identified as high risk and
they were asked to participate in the study. A
sample of African-American women who were
low in social support was identified. In this
study, the control group had a LBW rate of 22.4
percent, indicating that screening women at risk
for LBW was effectively achieved. The intervention group had a LBW rate of 9.1 percent,
192
suggesting that the social support intervention
for pregnant women with inadequate social
support can be effective in preventing LBW.
Most interventions have not
matched interventions to risks
Women with medical risk factors for LBW
require medical interventions; those with
psychosocial risk factors require psychosocial
interventions. Many women with known medical
risk factors have adequate social support and do
not need psychosocial interventions. In eight
out of twelve studies, participants were selected
based on medical, socioeconomic and behavioral risk factors, but not psychological risk
factors. Perhaps it is not surprising that offering
social support to these women was not effective
in reducing LBW rates.
In fact, social support may not always be
welcomed by participants. Developments in
social support theory clearly distinguish two
constructs of social support: received or
provided social support (also called enacted
social support), and perceived social support
(whether enacted or not) (Herbert & DunkelSchetter, 1992; Lakey & Lutz, 1996). Just
because the social support is received does not
mean it is accepted. In one study (Spencer et al.,
1989), over half the women randomly assigned
to a family social worker group refused to
participate in the study. It is also increasingly
recognized that social support is not always
helpful (Herbert & Dunkel-Schetter, 1992).
Providing social support inappropriately can
have an intrusive effect (Karfmacher, Kitzman,
& Olds, 1998), or negative effects of other kinds
(Dunkel-Schetter, Blasband, Feinstein, &
Herbert, 1992). For example, one study found
that reliance on peer networks for social support
was associated with higher distress (Reibstein,
1981). In another study involving family
members in home visits and helping them
become more bonded to the pregnancy (Graham
et al., 1992), few family members actually
participated in the study. These problems have
led researchers to rethink for whom, for what
and how should social support be provided.
In the only effective intervention trial,
Norbeck et al. (1996) designed an intervention
to provide emotional support based on focus
group results. The intervention consisted of four
standardized face-to-face individual sessions
02 Qian Lu (bc/d)
21/1/05
3:40 pm
Page 193
LU ET AL.: LEARNING FROM SUCCESS AND FAILURE
that followed written protocols. The pregnant
women were instructed to focus on three
problem areas and three successful areas in their
lives. Participants were guided to find meaning
in their pregnancy, to increase self-esteem and
improve relationships. This type of emotional
support was found to be effective in reducing
LBW rates among African-American women at
risk for low emotional support. The trial was
successful in part because the intervention
matched the risks.
Most interventions did not test
process variables
Among the 12 studies reviewed here, only one
tested process variables (Klerman et al., 2000).
Two other studies included measurement of
process variables in the study design, but did not
include them in the analyses (Olds et al., 1986;
Villar et al., 1992). Without testing process
variables, it is difficult to determine which
components of the intervention worked and
which failed (Stevens-Simon & Orleans, 1999).
The only pitfall in the design of the one successful intervention study (Norbeck et al., 1996) was
that process variables were not measured. Thus
it is not known whether the reduction in LBW
rate resulted from increased emotional support
as the predictive model hypothesized, or other
unmeasured processes.
Furthermore, as Stevens-Simon and Orleans
(1999) observed, when programs fail, it is often
unclear whether the failure is because a flawed
theory was implemented or because a good
theory was poorly implemented. The only intervention that tested process variables (Klerman
et al., 2000) failed to decrease LBW. However,
because the study tested process variables, it
still provided valuable information about why
the intervention failed. In this study, 629
women were randomized to receive usual or
augmented prenatal care. Augmented care
included educationally oriented peer groups,
additional appointments, extended time with
clinicians and other supports. Participants in the
augmented care group showed more positive
perceptions of their prenatal care and greater
knowledge about their risk factors. However,
there was no difference in LBW rates between
the two groups. Thus, the results did not support
the theory that perceived helpfulness or
increased knowledge (at least as manipulated)
decreased the rate of LBW. Future psychosocial
interventions should include measurement of
process variables in the study design.
Conclusion
Our review suggests that it may be premature to
conclude that psychosocial interventions are
ineffective in preventing LBW. Most intervention trials are flawed in their study design.
Thus before we can answer whether these interventions worked or not, we need to address how
to conduct high quality intervention studies.
Future intervention research should aim to: (1)
construct a predictive model a priori to guide
the intervention design; (2) develop an effective
risk screening based on modifiable psychosocial
risk factors, and tailor the psychosocial intervention to reduce these risk factors; and (3) test
process as well as outcome variables.
References
Aiken, L. S. (1999). Understanding how intervention
works: Combining research design and mediational
analysis. Presidential Address to Division 5, Evaluation, Measurement, and Statistics, of the American
Psychological Association, Boston, MA.
Aiken, L. S., West, S. G., Woodward, C. K., Reno,
R. R., & Reynolds, K. D. (1994). Increasing
screening mammography in asymptomatic women:
Evaluation of a second-generation, theory-based
program. Health Psychology, 13(6), 526–538.
Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator variable distinction in social psychological research: Conceptual, strategic and statistical
considerations. Journal of Personality and Social
Psychology, 51(6), 1173–1182.
Drotar, D. (2002). Enhancing reviews of psychological
treatments with pediatric populations: Thoughts on
next steps. Journal of Pediatric Psychology, 27(2),
167–176.
Dunkel-Schetter, C., Blasband, D., Feinstein, L. G., &
Herbert, T. B. (1992). Elements of supportive
interactions: When are attempts to help effective?
In S. Spacapan & S. Oskamp (Eds.), Helping and
being helped in the real world (pp. 83–114). Newbury
Park, CA: Sage Publications.
Fisher, J. D., Fisher, W. A., Misovich, S. J., Kimble.
D. L., & Malloy, T. E. (1996). Changing AIDS risk
behavior: Effects of an intervention emphasizing
AIDS risk reduction information, motivation, and
behavioral skills in a college student population.
Health Psychology, 15(2), 114–123.
193
02 Qian Lu (bc/d)
21/1/05
3:40 pm
Page 194
JOURNAL OF HEALTH PSYCHOLOGY 10(2)
Graham, A. V., Frank, S. H., Zyzanski, S. J., Kitson,
G. C., & Reeb, K. G. (1992). A clinical trial to
reduce the rate of low birth weight in an inner-city
black population. Family Medicine, 24(6), 439–446.
Heins, H. C., Webster, N. N., McCarthy, B. J., &
Melvin, C. E. (1990). A randomized trial of nursemidwifery prenatal care to reduce low birth weight.
Obstetrics and Gynecology, 75, 341–345.
Herbert, T. L., & Dunkel-Schetter, C. (1992).
Negative social reactions to victims: An overview of
responses and their determinants. In L. Montada,
S. H. Filipp, & M. J. Lerner (Eds.), Life crises and
experiences of loss in adulthood (pp. 497–518).
Hillsdale, NJ: Lawrence Erlbaum.
Hodnett, E. S., & Fredericks, S. (2003). Support during
pregnancy for women at increased risk of low birthweight babies. Cochrane Database Syst. Rev., No. 3,
CD000198.
Hoffman, S., & Hatch, M. C. (1996). Stress, social
support and pregnancy outcome: A reassessment
based on recent research. Pediatric Perinatal
Epidemiology, 10, 380–405.
Hudson, W. (1982). The clinical measurement package:
A field manual. Homewood, IL: Dorsey Press.
Karfmacher, J., Kitzman, H., & Olds, D. (1998). Intervention process as predictors of outcomes in a
preventive home-visitation program. Journal of
Community Psychology, 26, 49–64.
Kitzman, H., Olds, D. L., Henderson, C. R., Jr, Hanks,
C., Cole, R., Tatelbaum, R., McConnochie, K. M.,
Sidora, K., Luckey, D. W., Shaver, D., Engelhardt,
K., James, D., & Barnard, K. (1997). Effect of
prenatal and infancy home visitation by nurses on
pregnancy outcomes, childhood injuries, and repeat
childbearing: A randomized controlled trial. Journal
of American Medical Association, 278, 644–652.
Klerman, L. V., Ramey, S. L., Goldenberg, R. L.,
Marbury, S., Hou, J., & Cliver, S. P. (2000). A
randomized trial of augmented prenatal care for
multiple-risk, Medicaid-eligible African American
women. American Journal of Public Health, 91,
105–111.
Lakey, B., & Lutz, C. J. (1996). Social support and
preventive and therapeutic interventions. In G. R.
Pierce, B. R. Sarason, & I. G. Sarason (Eds.),
Handbook of social support and the family
(pp. 435–465). Plenum series on stress and coping.
New York: Plenum Press.
Lobel, M., Dunkel-Schetter, C., & Scrimshaw, S. C.
(1992). Prenatal maternal stress and prematurity: A
prospective study of socioeconomically disadvantaged women. Health Psychology, 11(1), 32–40.
Lu, M. C., Tache, V., Alexander, G. R., Kotelchuck,
M., & Halfon, N. (2003). Preventing low birth weight:
Is prenatal care the answer? Journal of Maternal
Fetal and Neonatal Medicine, 13(6), 362–380.
194
Mathews, T. J., Menacker, F., & MacDorman, M. F.
(2003). Infant mortality statistics from the 2001
period linked birth/infant death data set. National
Vital Statistics Report, 52(2), 1–28.
McLaughlin, F. J., Altemeier, W. A., Christensen, M.
J., Sherrod, K. B., Dietrick, M. S., & Stern, D. T.
(1992). Randomized trial of comprehensive prenatal
care for low-income women: Effect on infant birth
weight. Pediatrics, 89(1), 128–132.
Moore, M. L., Meis, P. J., Ernest, J. M., Wells, H. B.,
Zaccaro, D. J., & Terrell, T. (1998). A randomized
trial of nurse intervention to reduce preterm and
low birth weight births. Obstetrics and Gynecology,
91, 656–661.
Norbeck, J. S., & Anderson, N. J. (1989). Psychosocial
predictors of pregnancy outcomes in low-income
black, Hispanic, and white women. Nursing
Research, 38, 204.
Norbeck, J. S., Dejoseph, J. F., & Smith, R. T. (1996).
A randomized trial of an empirically-derived social
support intervention to prevent low birth weight
among African American women. Social Science
and Medicine, 43, 947–954.
Oakley, A., Rajan, L., & Grant, A. (1990). Social
support and pregnancy outcome. British Journal of
Obstetrics and Gynecology, 97, 155–162.
Olds, D. L., Henderson, C. R., Tatelbaum, R., &
Chamberlin, R. (1986). Improving the delivery of
prenatal care and outcomes of pregnancy: A
randomized trial of nurse home visitation. Pediatrics,
77, 16–28.
Paarlberg, K. M., Vingerhoets, A. J., Passchier, J.,
Dekker, G. A., & van Geijn, H. P. (1995). Psychosocial factors and pregnancy outcome: A review
with emphasis on methodological issues. Journal of
Psychosomatic Research, 39(5), 563–595.
Reibstein, J. (1981). Adjustment to the maternal role
in mothers leaving careers: The impact of their
interaction with role colleagues. Dissertation,
University of Chicago.
Rothberg, A. D., & Lits, B. (1991). Psychosocial
support for maternal stress during pregnancy:
Effect on birth weight. American Journal of Obstetrics and Gynecology, August, 403–407.
Smilkstein, G. (1978). The family APGAR: A
proposal for a family function test and its use by
physicians. Journal of Family Practice, 6, 1231–1236.
Spencer, B., Thomas, H., & Morris, J. (1989). A
randomized controlled trial of the provision of a
social support service during pregnancy: The South
Manchester Family Worker Project. British Journal
of Obstetrics and Gynecology, 96, 281–288.
Stevens-Simon, C., & Orleans, M. (1999). Low-birth
weight prevention programs: The enigma of failure.
Birth, 26, 184–191.
Villar, J., Farnot, U., Barros, F., Victora, C., Langer,
02 Qian Lu (bc/d)
21/1/05
3:40 pm
Page 195
LU ET AL.: LEARNING FROM SUCCESS AND FAILURE
A., & Belizan, J. (1992). A randomized trial of
psychosocial support during high-risk pregnancies.
New England Journal of Medicine, 327, 1266–1271.
West, S. G., & Aiken, L. S. (1997). Toward understanding individual effects in multicoponent prevention programs: Design and analysis strategies. In
K. J. Bryant, M. Windle, & S. G. West (Eds.),
The science of prevention: Methodological advances
from alcohol and substance abuse research
(pp. 167–209). Washington, DC: American Psychological Association.
195