Journal of Behavioral Medicine
J Behav Med (2012) 35:538-547
The impact of emotional well-being on
long-term recovery and survival in physical
illness: a meta-analysis
Sanne M.A.Lamers, Linda Bolier,
Gerben J.Westerhof, Filip Smit & Ernst
Your article is published under the Creative
Commons Attribution license which allows
users to read, copy, distribute and make
derivative works, as long as the author of
the original work is cited. You may self-
archive this article on your own website, an
institutional repository or funder’s repository
and make it publicly available immediately.
The impact of emotional well-being on long-term recovery
and survival in physical illness: a meta-analysis
Sanne M. A. Lamers •Linda Bolier •Gerben J. Westerhof •
Filip Smit •Ernst T. Bohlmeijer
Received: December 17, 2010 / Accepted: September 2, 2011 / Published online: September 15, 2011
ÓThe Author(s) 2011. This article is published with open access at Springerlink.com
Abstract This meta-analysis synthesized studies on emo-
tional well-being as predictor of the prognosis of physical
illness, while in addition evaluating the impact of putative
moderators, namely constructs of well-being, health-related
outcome, year of publication, follow-up time and methodo-
logical quality of the included studies. The search in refer-
ence lists and electronic databases (Medline and PsycInfo)
identiﬁed 17 eligible studies examining the impact of general
well-being, positive affect and life satisfaction on recovery
and survival in physically ill patients. Meta-analytically
combining these studies revealed a Likelihood Ratio of 1.14,
indicating a small but signiﬁcant effect. Higher levels of
emotional well-being are beneﬁcial for recovery and survival
in physically ill patients. The ﬁndings show that emotional
well-being predicts long-term prognosis of physical illness.
This suggests that enhancement of emotional well-being
may improve the prognosis of physical illness, which should
be investigated by future research.
Keywords Meta-analysis Emotional well-being
Recovery Survival Prognosis Physical illness
This meta-analysis investigates emotional well-being as a
predictor of the prognosis of physical illness. We deﬁne
emotional well-being from a positive perspective, not as
the mere absence of symptoms of psychopathology. Cur-
rently, most studies on the relation between mental and
physical health investigated the presence or absence of
psychopathology. These studies show that psychopathol-
ogy is related to the course and severity of several physical
diseases. For example, depression is associated with
increased osteoporosis (Michelson et al., 1996), coronary
heart disease (Glassman & Shapiro, 1998), diabetes com-
plications (De Groot et al., 2001), cancer incidence, pro-
gression (Spiegel & Giese-Davis, 2003) and cancer
mortality (Satin et al., 2009), and anxiety may inﬂuence the
development of coronary heart disease (Kubzansky &
By contrast, well-being may play an additional protec-
tive role in the course of physical diseases. After all, there
is accumulating evidence that psychopathology and well-
being are more than merely opposite poles of the same
dimension (Huppert & Whittington, 2003; Keyes, 2005,
2007; Lamers et al., 2011; Watson & Tellegen, 1985), and
both well-being and mental disorders may have indepen-
dent impacts on physical health. To date, six reviews of the
literature synthesized effects of well-being on physical
health (Chida & Steptoe, 2008; Diener & Chan, 2011;
Howell et al., 2007; Lyubomirsky et al., 2005; Pressman &
Cohen, 2005; Veenhoven, 2008). In general, the conclu-
sions are favorable with well-being being positively asso-
ciated to better health (Diener & Chan, 2011; Howell et al.,
2007; Lyubomirsky et al., 2005), reduced risk of illness and
injury (Pressman & Cohen, 2005), and lower mortality
rates (Chida & Steptoe, 2008; Pressman & Cohen, 2005;
Veenhoven, 2008). In samples of healthy people, the
results of these studies clearly point towards the positive
effects of well-being on physical health. However, results
appear to be mixed in physically ill populations.
S. M. A. Lamers (&)G. J. Westerhof
E. T. Bohlmeijer
Psychology, Health and Technology, University of Twente,
P.O. Box 217, 7500 AE Enschede, The Netherlands
L. Bolier F. Smit
Trimbos Institute, Utrecht, The Netherlands
J Behav Med (2012) 35:538–547
To illustrate, Howell et al. (2007) found positive effects
of well-being on physical health for both healthy and dis-
eased populations, although results differed across health
outcomes. The ﬁndings suggest that well-being may
enhance physical functioning in healthy adults and improve
management of symptoms in diseased adults. For example,
the likelihood of longevity increases for individuals with
high well-being compared to those with low well-being, and
this survival rate even increases 10% for individuals with
chronic diseases who report high versus low well-being. The
meta-analysis of Chida and Steptoe (2008) also shows pro-
tective effects of well-being on survival in diseased popu-
lations with renal failure and HIV. Even though Howell et al.
(2007) and Chida and Steptoe (2008) show that well-being
generally is related to better physical health in diseased
adults, Diener and Chan (2011), Pressman and Cohen
(2005), and Veenhoven (2008) report otherwise. Diener and
Chan (2011) conclude that ﬁndings with respect to diseased
populations are mixed. Although Pressman and Cohen
(2005) and Veenhoven (2008) state that there is too little
consistency in the data to draw robust conclusions, both
reviews suggest that there may be no effects or even adverse
effects of well-being on physical health. In general, the
pattern of research ﬁndings seems to point towards positive
effects or no effects in relatively mildly diseased adults,
where adherence to medication and behavioral factors such
as physical exercise could play a role, and negative effects in
severely diseased adults with high short-term mortality rates
(Pressman & Cohen, 2005; Veenhoven, 2008).
In sum, the existing reviews produce inconsistent evi-
dence with respect to well-being as a predictor of physical
health in diseased populations. Conclusions across healthy
and diseased populations differ, because the outcomes
differ as well. In healthy individuals, the desirable health
outcome is to stay healthy and to reduce mortality and the
development of physical illness. Individuals with physical
diseases already experience a diminished physical health,
resulting in a different set of aims, such as decreasing
symptom severity, preventing worsening of disease, and
increasing survival rates.
This meta-analysis will focus on physically diseased
patients, aiming to prospectively study the effects of
emotional well-being on the prognosis of physical disease.
The objective is to broadly investigate the prognosis,
including survival, disease progress, recovery, and func-
tional status. In addition, this systematic review will
investigate emotional well-being, deﬁned in the hedonic
tradition of well-being research (Diener et al., 1999). In
this research tradition, emotional well-being consists of an
affective component, concentrating on positive emotions
such as feelings of happiness, and a cognitive component,
concentrating on evaluations of life such as life satisfac-
tion. The previous literature reviews applied diverse deﬁ-
nitions and terminology of well-being, investigating
positive emotions (Pressman & Cohen, 2005; Veenhoven,
2008), positive emotions and positive dispositions such as
optimism and sense of humor (Chida & Steptoe, 2008;
Diener & Chan, 2011), or all positive psychological con-
structs (Howell et al., 2007). Moreover, several of these
reviews included studies which measured quality of life by
items on physical health and functioning (Howell et al.,
2007). Other studies used positive affect adjectives such as
active and energetic (Pressman & Cohen, 2005). These
items might measure physical health instead of well-being.
Thus to avoid confounding, this meta-analysis will employ
a strict and narrow focus on emotional well-being, and in
doing so will try to avoid contamination.
To further unravel the inconsistencies observed in
reviewed studies, this systematic review will apply meta-
analytic moderator analyses to evaluate how different
constructs of well-being, health-related outcome, year of
publication, follow-up time and sample size introduce their
own impact on outcome. Moreover, the methodological
quality of the included studies will be assessed and added
as a potential moderator, since effect sizes might be smaller
in high-quality studies than in other studies (Cuijpers et al.,
In sum, this meta-analysis will synthesize evidence that is
drawn from prospective studies on the relationship between
emotional well-being and the prognosis of physical illness,
in physically diseased samples across a range of health
outcomes. In addition, our study will encompass quality
assessment of the primary studies and we will employ meta-
analytical techniques such as meta-regression and meta-
analytic moderator analyses. The previous reviews of the
literature (Chida & Steptoe, 2008; Diener & Chan, 2011;
Howell et al., 2007; Lyubomirsky et al., 2005; Pressman &
Cohen, 2005; Veenhoven, 2008) included several of these
aims, but none of them combined all aspects into a single
systematic literature review. Since the research ﬁeld of
positive psychology is growing rapidly, this review will also
include several new studies on the relation between emo-
tional well-being and the prognosis of physical illnesses.
Selection of studies
Studies were included if they reported on emotional well-
being or aspects of emotional well-being and on the
prognosis of physical illness, aiming to evaluate the pro-
spective effects of well-being on the prognosis. Studies
J Behav Med (2012) 35:538–547 539
were excluded when (1) the study design was not pro-
spective; (2) emotional well-being was not measured (e.g.,
emotional well-being was measured otherwise than the
presence of general well-being, positive affect and/or life
satisfaction, emotional well-being was part of a composite
index, or psychopathology was examined as indicator of
well-being); (3) the study population was physically heal-
thy, mentally disordered, or consisted of institutionalized
elderly; (4) the paper included insufﬁcient information for
data extraction required for meta-analysis.
First, we searched the reference lists of the literature
reviews of Chida and Steptoe (2008), Diener and Chan
(2011), Howell et al. (2007), Lyubomirsky et al. (2005),
Pressman and Cohen (2005), and Veenhoven (2008) for
studies ﬁtting the inclusion criteria. Second, a systematic
search was performed in two electronic databases, Medline
and PsycInfo, up to March 2011. The main search strategy
was based on two key components: emotional well-being
and prognosis of physical illness. Terms on both compo-
nents were searched in title, abstract and keywords. Emo-
tional well-being included the following terms of which at
least one had to be present: (well-being) or (wellbeing) or
(happiness) or (happy) or (life satisfaction) or (positive
affect) or (positive mood) or (positive emotion*). In
addition, at least one term on prognosis of physical illness
had to be present. With respect to prognosis, we were
mainly interested in recovery outcomes, using terms as
functional status, health, and survival. However, since
recovery outcomes were not always explicitly mentioned,
we also included search terms on recovery processes
(recovery, rehabilitation, surgery, surgical, post-operative,
postsurgical, morbidity, remission, convalescence), general
terms of physical diseases (patient, disease, illness, pain,
surviv*, mortality, injury, fracture, infarction) and terms on
speciﬁc diseases (cancer, tumor, diabetes, arthritis, osteo-
arthritis, ﬁbromyalgia, arthrosis, heart failure, angina, car-
diac, cardiovascular, myocardial, coronary, thrombosis,
stroke, cardiovascular accident, COPD, lung disease,
bronchitis, aids, HIV). Only one of the search terms of
prognosis of physical illness had to be present.
We searched for peer-reviewed studies in the English
language with no limitations on the year in which the study
was published. To minimize the presence of publication
bias we also searched for dissertations. Furthermore, we
cross-checked the reference lists of included studies for
additional eligible studies. Potentially eligible studies were
independently selected by two reviewers (SL and LB) in
two phases. In the ﬁrst phase, selection was based on title
and abstract, and in the second phase on the full-text paper.
All studies evaluated as potentially eligible by at least one
of the reviewers in the ﬁrst selection phase, were evaluated
in the second selection phase. In the second phase, dis-
agreements between both independent reviewers were
resolved by consensus.
Our search revealed 17 eligible studies. The ﬂow diagram
of the study selection is shown in Fig. 1. Searching the
reference lists of the literature reviews (Chida & Steptoe,
2008; Diener & Chan, 2011; Howell et al., 2007; Lyubo-
mirsky et al., 2005; Pressman & Cohen, 2005; Veenhoven,
2008) and searching databases revealed in total 2,901
records. After exclusion (see Fig. 1), 17 studies were
included in the meta-analysis. Of these studies, 6 studies
were identiﬁed by searching the reference lists and 11 by
the electronic search, thus adding new studies to the pre-
Table 1shows an overview of the included studies. All
eligible studies were peer-reviewed articles. The study
populations were diverse, including heart and vascular
diseases (n=6), cancer (n=1), renal disease (n=1),
spinal cord injury (n=1), HIV (n=1), diabetes (n=1),
arthritis (n=1), stroke (n=1), hip fracture (n=1),
respiratory disorder (n=1), general acute events, includ-
ing stroke, hip fracture and heart attack (n=1), and gen-
eral medical patients (n=1). The sample sizes ranged
from 44 to 5,025 (M=749.7; SD =1139.5). Three types
of well-being constructs were extracted: general well-being
(n=1), positive affect (n=13), and life satisfaction
(n=3). The studies measured general well-being using the
WHO-5 well-being index (n=1; Heun et al., 1999),
positive affect using the subscale Positive affect of the
Center for Epidemiological Studies Depression Scale
(n=7; Radloff, 1977), the subscale Positive affect of the
Hospital Anxiety and Depression Scale (n=2; Herrmann,
1997), the Mood Adjective Check List (n=2; Nowlis,
1965), and the Global Mood Scale (n=2; Denollet, 1993).
Life satisfaction was measured by the Satisfaction With
Life Scale (n=1; Diener et al., 1985), the MOS short form
general health survey (n=1; Stewart et al., 1988), and the
Life Situation Questionnaire (n=1; Krause, 1992).
The type of outcome measures included functional sta-
tus (n=6), health status (n=1), and survival (n=10).
We combined functional status and health status as
recovery outcomes. Functional status was measured by the
Duke Activity Status Index (n=1; Hlatky et al., 1989),
(modiﬁed version of the) Katz’s Activities of Daily Living
scale (n=2; Katz et al., 1963), the Inpatient Rehabilita-
tion Facilities-Patient Assessment Instrument (n=1;
Ottenbacher et al., 1996), the EuroQol (Euroqol group,
1990), and by measuring usual walking speed, rapid
walking speed and chair stands (n=1). Of the 10 studies
540 J Behav Med (2012) 35:538–547
measuring survival as outcome, 9 studies measured sur-
vival status (alive or deceased at follow-up) and 1 study
measured survival time. The studies reported hazard ratios
(n=6), risk ratios (n=3), odds ratios (n=4), regression
coefﬁcients (n=3) or means (n=1). The follow-up time
ranged from 3 months to 11 years with a mean of
4.47 years (SD =3.93), and the papers were published
between 1996 and 2009.
Information was extracted on study design, type of study
population (e.g., cancer patients), sample size, type of well-
being construct, type of outcome measures, and the study’s
outcome measure. For each paper, we extracted the rele-
vant and most reliable outcome which was most com-
pletely adjusted for potential confounders, such that we
obtained a single outcome per paper. For one study (Ver-
steeg et al., 2009) we performed a meta-analysis to syn-
thesize three odds ratios on mobility, self-care, and
activities, into a single odds ratio. Moreover, we extracted
the results based on baseline emotional well-being instead
of change in emotional well-being over time. For one study
(Moskowitz, 2003), the result based on multiple measure-
ments of well-being was extracted, since the study did not
report results based on baseline emotional well-being.
When a paper included insufﬁcient information for data
extraction required for meta-analysis, we contacted the
authors for additional information.
Two reviewers (SL and LB) independently assessed
methodological quality of the included studies, using a
protocol based on the quality checklist for observational
studies of Wong et al. (2008). We adapted the checklist to
our study aims into a checklist consisting of ﬁve quality
criteria on external validity, response rate, reliability,
control for confounding demographic variables, and con-
trol for confounding health variables. For the studies on
recovery we included an additional item on the objective-
ness of the recovery measurements (i.e., self-report versus
laboratory test). Each criterion was rated as 0 (study does
not meet criterion) or 1 (study meets criterion). The
interrater reliability was 89.1%. The overall quality of the
study was assessed by dividing the total score by the total
number of applicable items, resulting in a quality score
between 0 and 1.00.
The quality of the studies ranged from 0.50 to 1.00
(M=0.74; SD =0.17). Three of the studies met all
quality criteria. In ten studies the reliability of the scale
measuring emotional well-being was not reported (n=1)
or Cronbachs alpha was lower than .60 (n=9), mainly
because well-being was measured by positive affect sub-
scales from depression and anxiety questionnaires. The
criterion assessing whether course of disease was measured
studies (n = 2,934)
Records identified through
reference list searching
(n = 67)
Additional records identified
through database searching
(n = 3,190) Dissertations
(n = 256)
Records after duplicates removed
(Medline, PsycInfo and reference lists)
(n = 2,901)
(n = 2,901)
(n = 2,783)
Full-text articles excluded
assessed for eligibility
(n = 118)
(n = 101)
1. Emotional well-being was not
measured (68 studies)
2. The study design was not
prospective (14 studies)
3. The study population was not
Studies included in
(n = 17)
4. The paper included insufficient
information for data extraction
required for meta-analysis (10
Fig. 1 Flow chart
J Behav Med (2012) 35:538–547 541
Table 1 Descriptives of the studies on emotional well-being as predictor of the course of physical disease
1.00 6 Chronic heart disease
Survival 1.024 (1.005–1.042) +
Brown et al.
0.80 10 Cancer (N=205) Positive affect
Survival 0.990 (0.938–1.045) 0
0.67 3 Coronary artery disease,
age 60+ (N=948)
1.609 (1.039–2.492) +
Denollet et al.
0.80 2 Coronary artery disease
Survival 2.550 (1.479–4.397) +
Fisher et al.
0.67 2 Arthritis, age 65+
Functional status (ADL) 1.099 (1.024–1.181) +
Fredman et al.
0.50 2 Hip fracture, age 65+
Functional status: usual
and rapid walking
speed, chair stands
2.700 (1.096–6.654) 0
Kimmel et al.
1.00 4 Hemodialysis patients
Survival 1.205 (0.960–1.513) 0
Konstam et al.
0.80 3 Congestive heart failure
Survival 0.949 (0.899–1.001) 0
Krause et al.
1.00 11 Spinal cord injury
Survival 1.990 (1.373–2.885) +
0.60 10.8 HIV + patients (N=407) Positive affect
Survival 1.163 (1.042–1.299) +
0.80 10 Diabetic patients
Survival 1.111 (0.962–1.284) 0
Olofson et al.
0.60 8 Chronic alveolar
Survival 1.961 (0.901–4.167) 0
Ostir et al.
0.50 1 Acute events (stroke, heart
attack or hip fracture),
age 65+ (N=240)
Functional status (ADL) 2.700 (1.096–6.653) +
Ostir et al.
0.50 0.25 Stroke, age 55+ (N=823) Positive affect
Pelle et al.
0.83 1 Chronic heart failure
Health status (HCS) 0.865 (0.603–1.241) 0
0.80 1 Patients of the general
medical ward (N=575)
(HADS; 1 item
Survival 1.400 (1.016–1.930) +
Versteeg et al.
0.83 1 Coronary artery disease
1.031 (0.649–1.667) 0
Range from 0.00 (low quality) to 1.00 (high quality), based on external validity, response rate, reliability, control for confounding demographic
variables, control for confounding health variables, and objectiveness of the recovery outcomes (not applicable for studies on survival)
Zung =Zung self-rating depression scale (SDS subscale well-being); WHO =WHO-5 well-being index; MACL Mood Adjective Check list,
HADS Hospital Anxiety and Depression Scale (subscale positive affect), SWLS Satisfaction with Life Scale; MOS MOS short form general health
survey, LSQ Life Situation Questionnaire (subscale), CES-D Center for Epidemiologic Studies Depression Scale (subscale positive affect), GMS
Global Mood Scale
DASI Duke Activity Status Inventory, ADL Activities of Daily Living Scale, IRF-PAI Inpatient Rehabilitation Facilities—Patient Assessment
Instrument, HCS Health Complaints Scale (subscale cardiac symptoms), EQ5D EuroQol-5D
+=Positive effect (PB.05); 0 =No effect (P[.05); -=Negative effect (PB.05)
542 J Behav Med (2012) 35:538–547
objectively (i.e., no self-report), was only applicable for the
studies measuring functional status or health status. Of
these studies (n=7), only one met this quality criterion.
We used the software Comprehensive Meta Analysis
(CMA) to meta-analytically combine study outcomes. For
each study, we extracted the hazard ratio, risk ratio or odds
ratio and its conﬁdence intervals. Regression coefﬁcients
and means were converted to odds ratios using CMA. We
combined the ratios, referring to the hazard ratios, risk
ratios and odds ratios as likelihood ratios (LR). When
necessary the ratio was inverted such that all LRs above 1
indicate a positive relationship of emotional well-being to
the prognosis of physical illness. The meta-analysis
included weighting of the study LRs by the inverse of the
standard errors, based on the conﬁdence intervals. With
small studies tending to have wider conﬁdence intervals
and large studies to have narrow conﬁdence intervals, the
conﬁdence interval reﬂects the precision of the LR. The LR
is considered statistically signiﬁcant if the conﬁdence
interval (95%) excludes the null value of 1.
A random-effects meta-analysis was performed, because
of the heterogeneity across the studies. The random-effects
method allows to assume that the studies are estimating
different but related effects, thus relacing the assumption
that all studies are replicas. In addition, the random-effects
model makes an adjustment to the study weights according
to the extent of heterogeneity (Deeks et al., 2008), which
translates into a broad 95% conﬁdence interval around the
pooled effect estimate.
We performed an overall analysis, as well as subgroup
analyses and meta-regression analyses to identify potential
moderators. In the subgroup analyses, we examined the
effects of emotional well-being on the prognosis of phys-
ical illness separately for positive affect and for each
measure of prognosis (survival; recovery). No subgroup
analyses were performed on overall well-being and life
satisfaction, since few studies measured these aspects of
emotional well-being (n=1 and 3, respectively). The
study population could not be split into more homogeneous
subgroups because of its (too large) diversity. In the meta-
regression analyses, we evaluated the potentially con-
founding relationship of sample size, the quality of the
studies, the follow-up length, and the publication year on
the relationship of interest: the impact of emotional well-
being on the prognosis of physical illness. To this end, we
used an unrestricted maximum likelihood mixed effects
regression. Moreover, we examined heterogeneity between
the studies by using the Q-test, indicating the probability of
heterogeneity, and the I
index, indicating the magnitude
of the heterogeneity. An I
between 0 and 30% was
considered as low, between 30 and 75% as moderate, and
between 75 and 100% as high heterogeneity (Deeks et al.,
Publication bias in the studies was evaluated using three
indices: the funnel plot, the Egger’s test of intercept and the
Rosenberg fail-safe number. The funnel plot is a graph of
effect size (LR) against sample size (N). When publication
bias is absent, the observed studies are expected to be
distributed symmetrically around the pooled effect size.
The Egger’s test of intercept is a correlation between study
precision (the inverse of the standard error) and the stan-
dardized effect (the effect size divided by its standard
error). The fail-safe number indicates the number of non-
signiﬁcant unpublished studies needed to reduce the overall
signiﬁcant effect to non-signiﬁcance (Sterne et al., 2008).
We used Rosenberg’s (2005) weighted method for calcu-
lating fail-safe numbers, where studies with small variance
are given higher weight than those with large variance. For
the reporting of this meta-analysis, we applied PRISMA
guidelines (Moher et al., 2009).
An overview of the 17 selected studies is presented in
Table 1. The studies investigated the prospective relation-
ship of general well-being, positive affect or life satisfac-
tion to survival or recovery. None of the studies reported
negative effects of emotional well-being on the prognosis
of physical illness. The results of the 17 studies and the
meta-analysis are presented in Fig. 2. Meta-analytically
summarizing the effects across the studies revealed an
overall likelihood ratio of 1.14 (P\0.001), indicating a
small but signiﬁcant effect of emotional well-being on the
course of physical disease. Since the studies were weigh-
ted, we also conducted a meta-analysis without the study of
Konstam et al. (1996), which is an outlier in sample size
(n=5,025) as reﬂected by its narrow conﬁdence interval.
Meta-analytically combining the remaining 16 studies
revealed a higher likelihood ratio of 1.18 (P\.001),
indicating the positive relation of emotional well-being to
recovery and survival is even stronger when excluding the
study with the highest weight. The 17 studies were heter-
ogeneous as the variability of the effect sizes is larger than
would be expected from sampling error alone. This high
heterogeneity indicated that variability across the primary
studies largely stems from systematic factors, such as the
type of studied well-being, type of outcome or differences
in methodological quality of the studies.
We performed two subgroup analyses to evaluate whe-
ther the effects differed for positive affect or course of
disease outcomes. When examining the effects separately
for positive affect, the 13 studies on positive affect
J Behav Med (2012) 35:538–547 543
revealed an LR of 1.22 (P\.001). Positive affect was
signiﬁcantly related to more survival and recovery. The
distribution of effect sizes within studies on positive affect
remained heterogeneous. Second, we performed a sub-
group analysis on type of outcome. We split the studies in
two groups, measuring recovery (n=7) or survival
(n=10). Meta-analytically combining the studies on
recovery and survival resulted in LRs of 1.39 (P=.02)
and 1.11 (P=.01), respectively. Emotional well-being
signiﬁcantly predicted both survival and recovery later in
time, with the strongest relation to recovery. Although the
LR is higher for recovery, the conﬁdence intervals show
that the ratio for recovery is estimated with less precision
than the ratio for survival. When taking type of outcome
into account as a moderator, the heterogeneity remained
moderate to large within studies on recovery and survival.
Moreover, we evaluated quality of the studies, follow-up
length, publication year, and sample size as moderators of
the relation between emotional well-being and course of
disease by performing three meta-regression analyses.
Results of the meta-regression were insigniﬁcant for study
quality (B=-.56; P=.24), follow-up length (B=-.01;
P=.53), publication year (B=.01; P=.59), and for
sample size (B=-.00; P=.23).
Finally, we evaluated publication bias. The funnel plot
indicated asymmetry, since the studies are mainly con-
centrated on the right side of the plot. The Egger’s test of
intercept (t=4.41; df =15; P\.001) also suggests that
bias exists. There is a signiﬁcant correlation between study
precision (the inverse of the standard error) and the stan-
dardized effect (the effect size divided by the standard
error). Moreover, the fail-safe number indicated that 2.4
non-signiﬁcant unpublished studies must be included in our
random-effects model to reduce the overall signiﬁcant
effect to non-signiﬁcance. The funnel plot, Egger’s test of
intercept and fail-safe number indicated the presence of
publication bias. To gain insight in the grey literature, we
searched the electronic databases for eligible dissertations.
Three dissertations were eligible (Caron, 1997; Hamilton,
1996; Ostir, 2001), but excluded because the studies were
already included in the meta-analysis (Ostir, 2001) or data
required for meta-analysis were unavailable despite con-
tacting the authors for additional information (Caron, 1997;
Hamilton, 1996). These three dissertations reported posi-
tive effects in the dissertation abstract, showing positive
effects of well-being on course of disease in unpublished
studies. Furthermore, given the novelty of the focus on
positive well-being in relation to physical health, it is
unlikely that many studies with negative ﬁndings are
unreported (Diener & Chan, 2011). However, we have to
interpret the results in our meta-analysis carefully, since
effects of well-being on course of disease might be over-
This meta-analysis synthesized studies on emotional well-
being as predictor of the prognosis of physical illness,
while in addition evaluating the impact of putative mod-
erators such as type of outcome. Although previous liter-
ature reviews included several of these study aims
(Lyubomirsky et al., 2005; Pressman & Cohen, 2005;
Howell et al., 2007; Chida & Steptoe, 2008; Veenhoven,
2008; Diener & Chan, 2011), this is the ﬁrst study to our
knowledge that combines all aspects into a single review.
Our literature search identiﬁed 17 eligible papers, of which
nine new studies in addition to the studies included in
earlier reviews. This shows that the research ﬁeld on the
relation of positive well-being to physical health is growing
Meta-analytically combining these studies showed that
positive emotional well-being is favorably related to the
prognosis of physical illness. Patients with higher baseline
Fig. 2 Forest plot
544 J Behav Med (2012) 35:538–547
levels of emotional well-being have better recovery and
survival rates than patients with low levels of emotional
well-being. Although the effect is small, it is remarkable
that emotional well-being at baseline has signiﬁcant effects
on physical health later in time, since the average follow-
up is approximately after 4 years. The effect size of the
relation between emotional well-being and course of dis-
ease is even similar for studies with short and long follow-
Subgroup analyses indicated that emotional well-being
is related to both survival and recovery. Moreover, positive
affect is beneﬁcial. Positive affect may inﬂuence immune
and cardiovascular systems directly by activating the
autonomic nervous system and the Hypothalamic–Pitui-
tary–Adrenal axis (HPA) thus buffering the impact of
stress. Positive affect has also an indirect favorable effect
by increasing health behavior and engagement in social
networks (Pressman & Cohen, 2005; Howell et al., 2007).
There were not enough studies to conduct a subgroup
analysis on life satisfaction (3 studies) and well-being (1
Recommendations for future research
It is important for future research to take into account that
the impact of emotional well-being on course of disease
might differ across health outcomes, well-being measures,
and study populations. As suggested in previous literature
reviews, effects might also differ across diseases (Chida
and Steptoe, 2008). In addition, Pressman and Cohen
(2005) and Veenhoven (2008) state that well-being has
beneﬁcial effects in relatively mildly diseased adults and
negative effects in severely diseased adults. For our future
understanding of the role of well-being in disease pro-
gression, it is highly relevant whether well-being has
similar effects in various disease populations. Unfortu-
nately, the study populations in the current meta-analysis
were too diverse to further investigate effects across dis-
Moreover, other population characteristics could play a
role in the relation between emotional well-being and
course of disease. Pressman and Cohen (2005) found that
positive affect was associated with lower mortality rates,
mainly in older community-residing adults. They suggest
that the association is possibly stronger in older partici-
pants. This meta-analysis could not unravel the effects of
age and health outcome, since the studies on survival
included patients of all ages, whereas studies on recovery
more often included only older patients. Additionally,
results could differ across gender. Brummett et al. (2009)
and Fisher et al. (2004) found that effects of well-being on
recovery were stronger for males than for females. In the
current meta-analysis, the patient populations were too
diverse to examine differential effects across well-being
and outcome measures, diseases, age and gender, but we
recommend investigating these aspects in future research.
Strengths and limitations
One of the strengths of this systematic review is the focus
on emotional well-being as the presence of well-being,
positive affect, or life satisfaction. We investigated positive
psychological aspects, whereas previous reviews also
included studies on quality of life which use items on
physical functioning and health. The focus on other aspects
than emotional well-being was our main criterion for
exclusion of studies. Aspects as vitality, energy, and opti-
mism might indirectly measure health (Pressman & Cohen,
2005). In addition, we included control for baseline health
status as a quality criterion, which was present in 14 of the
17 included studies.
Most studies used subscales from depression scales to
measure positive affect. For example, both the Hospital
Anxiety and Depression Scale (HADS) and the Center for
Epidemiologic Studies Depression Scale (CES-D) include
a subscale on positive affect. However, these question-
naires are designed to screen for depressive symptoms
rather than to measure positive affect, resulting in low
subscale reliability (Penninx, 2000). Additional studies
need to be conducted, using reliable questionnaires which
are designed to measure well-being, such as the Positive
and Negative Affect Scales (PANAS; Watson et al., 1988)
and Mental Health Continuum-Short Form (MHC-SF;
Lamers et al., 2011).
Although most studies used depression scales to measure
aspects of positive well-being, few studies included nega-
tive aspects of mental health such as psychopathology and
negative affect as a confounding variable, to evaluate the
unique effects of emotional well-being. Studies that did
evaluate the unique effects of positive emotional well-being
report positive results, such as Brummett et al. (2009).
In addition, the results from our meta-analysis have to
be interpreted carefully. First, the studies used different
covariates varying from baseline health characteristics to
demographic characteristics such as gender and age,
making the studies diverse. The high heterogeneity con-
ﬁrmed that the variability across the studies was larger than
would be expected from sampling error alone. Although
study quality, in which control for baseline health status
and for demographics were used as quality criteria, was not
related to the effect size of the study, we have to take the
high variability and diversity in covariates across studies
into account. Moreover, the results have to be interpreted
carefully because of potential publication bias. The effects
of emotional well-being on recovery and survival might be
J Behav Med (2012) 35:538–547 545
Conclusion and implications
Emotional well-being predicts long-term prognosis of
physical illness. Higher levels of emotional well-being are
beneﬁcial for recovery and survival rates in physically
diseased patients. Although the effects are small, the
ﬁndings are important. Recovery and survival are highly
relevant outcomes. Moreover, since physical diseases such
as coronary heart disease and cancer are highly prevalent,
small effects of emotional well-being on prognosis of
physical illness have a large impact in the population. In
addition, several psychological interventions are effective
in enhancing well-being, such as Acceptance and Com-
mitment Therapy (Fledderus et al., 2010) and well-being
therapy (Fava et al., 1998). By the enhancement of well-
being, these interventions might also improve recovery and
survival in physical illness. Future research should inves-
tigate effects of psychological interventions on the prog-
nosis of physical illness.
Open Access This article is distributed under the terms of the
Creative Commons Attribution Noncommercial License which per-
mits any noncommercial use, distribution, and reproduction in any
medium, provided the original author(s) and source are credited.
References marked with an asterisk [*] indicate studies
included in the meta-analysis.
*Birket-Smith, M., Hansen, B. H., Hanash, J. A., Hansen, J. F., &
Rasmussen, A. (2009). Mental disorders and general well-being
in cardiology outpatients—6 year survival. Journal of Psycho-
somatic Research, 67, 5–10.
*Brown, K. W., Levy, A. R., Rosberger, Z., & Edgar, L. (2003).
Psychological distress and cancer survival: A follow-up 10 years
after diagnosis. Psychosomatic Medicine, 65, 636–643.
*Brummett, B. H., Morey, M. C., Boyle, S. H., & Mark, D. B. (2009).
Prospective study of associations among positive emotion and
functional status in older patients with coronary artery disease.
Journal of Gerontology: Psychological Sciences, 64B, 461–469.
Caron, M. D. (1997). Psychological and environmental factors of
rehabilitation outcome in the elderly. Dissertation Abstracts
International, 57, 4335. (UMI No. 9638939).
Chida, Y., & Steptoe, A. (2008). Positive psychological well-being
and mortality: A quantitative review of prospective observa-
tional studies. Psychosomatic Medicine, 70, 741–756.
Cuijpers, P., Van Straten, A., Bohlmeijer, E., Hollon, S. D., &
Andersson, G. (2010). The effects of psychotherapy for adult
depression are overestimated: A meta-analysis of study quality
and effect size. Psychological Medicine, 40, 211–223.
De Groot, M., Anderson, R., Freedland, K. E., Clouse, R. E., &
Lustman, P. J. (2001). Association of depression and diabetes
complications: A meta-analysis. Psychosomatic Medicine, 63,
Deeks, J. J., Higgins, J. P. T., & Altman, D. G. (2008). Analysing data
and undertaking meta-analysis. In J. P. T. Higgins & S. Green
(Eds.), Cochrane handbook for systematic reviews of interven-
tions (pp. 243–297). Chichester: Wiley.
Denollet, J. (1993). Emotional distress and fatigue in coronary heart
disease: The global mood scale (GMS). Psychological Medicine,
*Denollet, J., Pedersen, S. S., Daemen, J., De Jaegere,P., Serruys, P. W.,
& van Domburg, R. T. (2008). Reduced positive affect (anhedo-
nia) predicts major clinical events following implantation of
coronary-artery stents. Journal of Internal Medicine, 263,
Diener, E., & Chan, M. Y. (2011). Happy people live longer:
Subjective well-being contributes to health and longevity.
Applied Psychology: Health and Well-being, 3, 1–43.
Diener, E., Emmons, R. A., Larsen, R. J., & Grifﬁn, S. (1985). The
satisfaction with life scale. Psychological Assessment, 5,
Diener, E., Suh, E. M., Lucas, R., & Smith, H. L. (1999). Subjective
well-being: Three decades of progress. Psychological Bulletin,
Euroqol group. (1990). A new facility for the measurement of health-
related quality of life. Health Policy, 16, 199–208.
Fava, G. A., Rafanelli, C., Cazzaro, M., Conti, S., & Grandi, S.
(1998). Well-being therapy. A novel psychotherapeutic approach
for residual symptoms of affective disorders. Psychological
Medicine, 28, 475–480.
*Fisher, M. N., Snih, S. A., Ostir, G. V., & Goodwin, J. S. (2004).
Positive affect and disability among older Mexican Americans
with arthritis. Arthritis and Rheumatism, 51, 34–39.
Fledderus, M., Bohlmeijer, E. T., Smit, F., & Westerhof, G. J. (2010).
Mental health promotion as a new goal in public mental health
care: A randomized controlled trial of an intervention enhancing
psychological ﬂexibility. American Journal of Public Health,
*Fredman, L., Hawkes, W. G., Black, S., Bertrand, R. M., &
Magaziner, J. (2006). Elderly patients with hip fracture with
positive affect have better functional recovery over 2 years.
Journal of the American Geriatrics Society, 54, 1074–1081.
Glassman, A. H., & Shapiro, P. A. (1998). Depression and the course
of coronary artery disease. American Journal of Psychiatry, 155,
Hamilton, A. B. (1996). Do psychosocial factors predict metastatic
cancer survival? A longitudinal assessment of breast, colon, and
lung cancer patients. Dissertation Abstracts International, 57,
4029. (UMI No. 9632697).
Herrmann, C. (1997). International experiences with the hospital
anxiety and depression scale: A review of validation data and
clinical results. Journal of Psychosomatic Research, 42,
Heun, R., Burkart, M., Maier, W., & Bech, P. (1999). Internal and
external validity of the WHO well-being scale in the elderly
general population. Acta Psychiatrica Scandinavica, 23, 56–62.
Hlatky, M. A., Boineau, R. E., Higginbotham, M. B., Lee, K. L.,
Mark, D. B., Califf, R. M., et al. (1989). A brief self-
administered questionnaire to determine functional capacity
(the Duke Activity Status Index). American Journal of Cardi-
ology, 64, 651–654.
Howell, R. T., Kern, M. L., & Lyubomirsky, S. (2007). Health
beneﬁts: Meta-analytically determining the impact of well-being
on objective health outcomes. Health Psychology Review, 1,
Huppert, F. A., & Whittington, J. E. (2003). Evidence for the
independence of positive and negative well-being: Implications
for quality of life assessment. British Journal of Health
Psychology, 8, 107–122.
Katz, S., Ford, A. B., Moskowitz, R. W., Jackson, B. A., & Jaffe, M. W.
(1963). The index of ADL: A standardized measure of biological
and psychosocial function. Journal of the American Medical
Association, 185, 914–919.
546 J Behav Med (2012) 35:538–547
Keyes, C. L. M. (2005). Mental illness and/or mental health?
Investigating axioms of the complete state model of health.
Journal of Consulting and Clinical Psychology, 73, 539–548.
Keyes, C. L. M. (2007). Promoting and protecting mental health as
ﬂourishing: A complementary strategy for improving national
mental health. American Psychologist, 62(2), 95–108.
*Kimmel, P. L., Peterson, R. A., Weihs, K. L., Simmens, S. J.,
Alleyne, S., Cruz, I., et al. (1998). Psychosocial factors,
behavioral compliance and survival in urban hemodialysis
patients. Kidney International, 54, 245–254.
*Konstam, V., Salem, D., Pouleur, H., Kostis, J., Gorkin, L.,
Shumaker, S., et al. (1996). Baseline quality of life as a
predictor of mortality and hospitalization in 5, 025 patients with
congestive heart failure. American Journal of Cardiology, 78,
Krause, J. S. (1992). Life satisfaction after spinal cord injury: A
descriptive study. Rehabilitation Psychology, 37, 61–70.
*Krause, J. S., Sternberg, M., Lottes, S., & Maides, J. (1997).
Mortality after spinal cord injury: An 11-year prospective study.
Archives of Physical Medicine and Rehabilitation, 78, 815–821.
Kubzansky, L. D., & Kawachi, I. (2000). Going to the heart of the
matter: Do negative emotions cause coronary heart disease?
Journal of Psychosomatic Research, 48, 323–337.
Lamers, S. M.A., Westerhof, G. J., Bohlmeijer, E. T., ten Klooster, P. M.,
& Keyes, C. L. M. (2011). Evaluating the psychometric properties
of the mental health continuum-short form. Journal of Clinical
Psychology, 67, 99–110.
Lyubomirsky, S., King, L., & Diener, E. (2005). The beneﬁts of
frequent positive affect: Does happiness lead to success?
Psychological Bulletin, 131, 803–855.
Michelson, D., Stratakis, M. D., Hill, L., Reynolds, J., Galliven, E.,
Chrousos, G., et al. (1996). Bone mineral density in women with
depression. The New England Journal of Medicine, 335,
Moher, D., Liberati, A., Tetzlaff, J., Altman, D. G., & The PRISMA
Group. (2009). Preferred reporting items for systematic reviews
and meta-analyses: The PRISMA statement. PLoS Medicine, 6,
*Moskowitz, J. T. (2003). Positive affect predicts lower risk of AIDS
mortality. Psychosomatic Medicine, 65, 620–626.
*Moskowitz, J. T., Epel, E. S., & Acree, M. (2008). Positive affect
uniquely predicts lower risk of mortality in people with diabetes.
Health Psychology, 27, 73–82.
Nowlis, V. (1965). Research with the mood adjective checklist.
In S. S. Tompkins & C. E. Izard (Eds.), Affect, cognition and
personality (pp. 352–389). New York: Springer.
*Olofson, J., Dellborg, C., Sullivan, M., Midgren, B., Caro, O., &
Bergman, B. (2009). Quality of life and palliation predict
survival in patients with chronic alveolar hypoventilation and
nocturnal ventilatory support. Quality of Life Research, 18,
Ostir, G. V. (2001). Emotional well-being and subsequent health,
functional ability and mortality in older non-Hispanic Whites,
Blacks and Mexican Americans. Dissertation Abstracts Interna-
tional, 61, 2865. (UMI No. 9981512).
*Ostir, G. V., Berges, I., Ottenbacher, M. E., Clow, A., &
Ottenbacher, K. J. (2008). Associations between positive emo-
tion and recovery of functional status following stroke. Psycho-
somatic Medicine, 70, 404–409.
*Ostir, G. V., Goodwin, J. S., Markides, K. S., Ottenbacher, K. J.,
Balfour, J., & Guralnik, J. M. (2002). Differential effects of
premorbid physical and emotional health on recovery from acute
events. Journal of the American Geriatrics Society, 50, 713–718.
Ottenbacher, K. J., Hsu, Y., Granger, C. V., & Fiedler, R. C. (1996).
The reliability of the functional independence measure: A
quantitative review. Archives of Physical Medicine and Reha-
bilitation, 77, 1226–1232.
*Pelle, A. J., Pedersen, S. S., Szabo, B. M., & Denollet, J. (2009).
Beyond type D personality: Reduced positive affect (anhedonia)
predicts imparied health status in chronic heart failure. Quality of
Life Research, 18, 689–698.
Penninx, B. (2000). A happy person, a healthy person? Journal of the
American Geriatrics Society, 48, 590–592.
Pressman, S. D., & Cohen, S. (2005). Does positive affect inﬂuence
health? Psychological Bulletin, 131, 925–971.
Radloff, L. S. (1977). The CES-D scale: A self report depression scale
for research in the general population. Applied Psychological
Measurement, 1, 385–401.
Rosenberg, M. S. (2005). The ﬁle-drawer problem revisited: A
general weighted method for calculating fail-safe numbers in
meta-analysis. Evolution, 59, 464–468.
Satin, J. R., Linden, W., & Phillips, M. J. (2009). Depression as a
predictor of disease progression and mortality in cancer patients.
Cancer, 115, 5349–5361.
*Scherer, M., & Hermann-Lingen, C. (2009). Single item on positive
affect is associated with 1-year survival in consecutive medical
inpatients. General Hospital Psychiatry, 31, 8–13.
Spiegel, D., & Giese-Davis, J. (2003). Depression and cancer:
Mechanisms and disease progression. Biological Psychiatry, 54,
Sterne, J. A. C., Egger, M., & Moher, D. (2008). Addressing reporting
biases. In J. P. T. Higgins & S. Green (Eds.), Cochrane
handbook for systematic reviews of interventions (pp. 297–333).
Stewart, A. L., Hays, R. D., & Ware, J. E. (1988). The MOS short
form general health survey: Reliability and validity in a patient
population. Medical Care, 26, 724–735.
Veenhoven, R. (2008). Healthy happiness: Effects of happiness on
physical health and the consequences for preventive health care.
Journal of Happiness Studies, 9, 449–469.
*Versteeg, H., Pedersen, S. S., Erdman, R. A. M., van Nierop, J.
W. I., de Jaegere, P., & Van Domburg, R. T. (2009). Negative
and positive affect are independently associated with patient-
reported health status following percutaneous coronary inter-
vention. Quality of Life Research, 18, 953–960.
Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and
validation of brief measures of positive and negative affect: The
PANAS scales. Journal of Personality and Social Psychology,
Watson, D., & Tellegen, A. (1985). Toward a consensual structure of
mood. Psychological Bulletin, 98, 219–235.
Wong, W. C. W., Cheung, C. S. K., & Hart, G. J. (2008).
Development of a quality assessment tool for systematic reviews
of observational studies of HIV prevalence in men having sex
with men and associated risk behaviours. Emerging Themes in
Epidemiology, 5, 23–26.
J Behav Med (2012) 35:538–547 547