Are we witnessing the decline effect in the Type D personality literature? What can
James C. Coynea,b,⁎, Jacob N. de Voogda,c
aHealth Psychology Section, Department of Health Sciences, University Medical Center Groningen, University of Groningen, The Netherlands
bDepartment of Psychiatry, Perelman School of Medicine of the University of Pennsylvania, USA
cDepartment of Pulmonary Rehabilitation, Center for Rehabilitation, University Medical Center Groningen, The Netherlands
a b s t r a c ta r t i c l ei n f o
Received 2 July 2012
received in revised form 18 September 2012
accepted 21 September 2012
After an unbroken series of positive, but underpowered studies seemed to demonstrate Type D personality
predicting mortality in cardiovascular disease patients, initial claims now appear at least exaggerated and
probably false. Larger studies with consistently null findings are accumulating. Conceptual, methodological,
and statistical issues can be raised concerning the construction of Type D personality as a categorical variable,
whether Type D is sufficiently distinct from other negative affect variables, and if it could be plausibly
assumed to predict mortality independent of depressive symptoms and known biomedical factors, including
disease severity. The existing literature concerning negative affect and health suggests a low likelihood of dis-
covering a new negative affect variable that independently predicts mortality better than its many rivals. The
apparent decline effect in the Type D literature is discussed in terms of the need to reduce the persistence of
false positive findings in the psychosomatic medicine literature, even while preserving a context allowing
risk-taking and discovery. Recommendations include greater transparency concerning research design and
analytic strategy; insistence on replication with larger samples before accepting “discoveries” from small
samples; reduced confirmatory bias; and availability of all relevant data. Such changes would take time to
implement, face practical difficulties, and run counter to established practices. An interim solution is for
readers to maintain a sense of pre-discovery probabilities, to be sensitized to the pervasiveness of the decline
effect, and to be skeptical of claims based on findings reaching significance in small-scale studies that have
not been independently replicated.
© 2012 Elsevier Inc. All rights reserved.
Many “discoveries” turnouttobeexaggeratedor simply false across
diverse research areas and disciplines. John Ioannidis  provoked con-
cern with his demonstration that of the 49 most cited clinical research
studies in major journals, replications of 34 had been attempted, and
41% of key findings had been refuted or shown to be substantially
diminished. In a subsequent paper  he offered evidence that “most
claimed research findings are false” and he and others [3–6] have
since identified some mechanisms by which discoveries appear in the
literature and then undergo a decline or refutation. Claims of break-
through discoveries frequently arise in small positive studies that
would seem to be too underpowered to detect an effect, and the inade-
quate sample size makes findings all the more attention grabbing.
Yet, portrayals in the literature of such “discoveries” ignore the larg-
er context of a strong confirmatory bias in published research papers,
with unknown numbers of negative findings not being published and
findings being declared discoveries simply because of having achieved
an arbitrary level of significance. Moreover, statistically significant
findings in small studies are necessarily always large because larger
effect sizes are needed to cross the higher threshold for significance.
of confirmatory bias, flexible rules of design, data analysis and reporting
[5,6], and significance chasing . There is a recurring decline effect 
in diverse literatures, usually occurring when discoveries declared on
the basis of small studies attract more resources or the attention of
new investigator groups with less of an investment in the discovery,
and attempted replications or extension of the original findings fail.
The rise and apparent decline of Type D personality
Are we now witnessing a decline effect in the literature concerning
Type D personality? The concept, which has been defined as the ten-
dency to experience negative emotions and to inhibit self-expression
for mortality in cardiovascular disease (CVD) patients. A preliminary
Journal of Psychosomatic Research 73 (2012) 401–407
⁎ Corresponding author at: Department of Psychiatry, Perelman School of Medicine
of the University of Pennsylvania, 3535 Market St. Rm 676, Philadelphia, PA 19104,
USA. Tel.: +1 215 662 7035; fax: +1 215 349 5067.
E-mail address: firstname.lastname@example.org (J.C. Coyne).
Contents lists available at SciVerse ScienceDirect
Journal of Psychosomatic Research
0022-3999/$ – see front matter © 2012 Elsevier Inc. All rights reserved.
entdistressed personalitytypeconstructed bycrosstabulatingdichoto-
mized continuous measures of trait anxiety and social inhibition,
selecting the high/high quadrant and compared it to the other three
. Low exercise tolerance, a previous MI, anterior site of the MI,
smoking, and age were entered into a logistic regression predicting 15
deaths and it was found that addition of the distressed personality
type added to the prediction. A subsequent Lancet study constructed a
Type D personality with different, but similar personality measures
and claimed that patients with Type D personality were four times
more likely be among the 21 dead during the observation period .
A later study claimed that patients with Type D personality were ten
times more likely to die after a cardiac transplant . Claims were also
made that its predictive value was independent of disease severity
and other biological variables. An unbroken series of five published
studies [8–12] found a significant relationship between Type D person-
ality and subsequent death.
at Tilburg University, with very small numbers of deaths being
study , followed by 6 , 4 , 6 , and 12  respectively (See
Table 1). These studies consistently found Type D to be associated with
tors of mortality in heart disease, even if the effect was present.
Observed magnitudes of effect were excessive relative to what could
be reasonably expected and, in most cases, far stronger than traditional
risk factors for mortality and recurrence, unless an extraordinary and
unprecedented psychological risk for mortality had been uncovered.
After the initial “discovery” of an effect on mortality in a sample
with as few deaths to explain as was the case in Type D studies, the
continued reliance on small studies implicitly committed investiga-
tors to the assumption that they were pursuing a replication of an
effect that is larger than many traditional biomedical risk factors. In
hindsight, this commitment should have been made explicit and
should have required elaborate justification beyond the initial finding
with a small sample.
Moreover, the statistical significance of results in these earlier
studies was typically vulnerable to reclassification or addition or sub-
traction of a few or even a single death. Observed results might well
be due to capitalization on chance or confirmatory bias allowed by
such factors as flexible extension or contraction of follow-up periods;
selective exclusion of some deaths; selective definition of endpoints
(cardiac-specific mortality versus all cause mortality versus mortality
included in a composite endpoint); selective inclusion of possible
covariates and exclusion of others, depending on the impact on
Type D significance and magnitude; or reporting of significance levels
only for adjusted, rather than unadjusted analyses. For example,
reports of these studies offered no indication that follow-up periods
and their stopping points were predetermined or whether follow-up
was stopped or extended, depending on whether a significant predic-
tion of mortality had been obtained.
Subsequently, the Tilburg group produced three studies that had
more deaths to explain and therefore a greater likelihood of reliably
detecting the value of Type D as a prognostic indicator for mortality.
The first of the investigator groups' larger studies attempted to pre-
dict 47 deaths, but the prognostic value of Type D personality did
not hold, once appropriate controls were introduced . The second
study attempted to predict 123 deaths among 641 heart failure pa-
tients, but Type D personality failed to be a significant prognostic in-
dicator in either bivariate or multivariate analyses with statistical
controls . The third study attempted to predict 187 deaths
among 1234 consecutive CVD patients receiving a percutaneous cor-
onary intervention and found no association between Type D person-
ality and mortality .
Recently, there have been several studies from outside the original
investigator group. The first one failed to find a prognostic value for
Type D, but was underpowered to do so, with only 11 deaths being
explained . Then, another two more studies from outside the
original investigator group were published, with larger numbers of
Studies assessing Type D and mortality in patients with cardiovascular diseases
Type D measureType D prevalenceUnadjusted effect size with
Adjusted effect size with
Denollet 1995 10515SI: heart patients psychological
questionnaire ; NA: trait STAI
median split SI≥12, NA≥40
SI: heart patients psychological
questionnaire; NA: trait STAI;
median split SI≥12, NA≥43
DS16; median split NA≥9, SI≥15
Denollet 1996 21021
Denollet 2000 319631%OR 11.65
Denollet 2006 3374DS16; median split A≥9, SI≥1529%
Pedersen 2007 35812DS14; NA/SI≥1030%HR 2.61
Denollet 2007 516DS14: NA/SI≥1029%2
Schiffer 2010 23247DS14; NA/SI≥1021%
Pelle 2010 641123DS14; NA/SI≥1020%21
Volz 2011 11111DS14; NA/SI≥1030%
Grande 2011 977172DS14; NA/SI≥1025%19
Coyne 2011 706192DS14; NA/SI≥1013%2
Damen 2012 1234187DS14; NA/SI≥1029%22
SI: social inhibition, NA: negative affectivity, STAI: State-Trait Anxiety Inventory, DS16: Type D scale 16-item version, DS14: Type D scale 14-item version, CI: confidence interval, nr:
not reported in original paper.
N.B.: length of follow-up is not reported because descriptive statistics were inconsistent.
J.C. Coyne, J.N. de Voogd / Journal of Psychosomatic Research 73 (2012) 401–407
deaths to be explained than most of the previous studies, but neither
or adjusted models, despite one having 172 deaths among 977 patients
with CVD  and the other study having 192 deaths at 18 months
follow-up of 706 HF patients .
Both of the studies relied on all cause, rather than cardiac-specific
on all cause mortality since cause of death is not reliably reported in
death certificates and because of the unreliability of judgments that
death is cardiac specific in an elderly population with numerous
 had a shorter follow-up period (18 months) than earlier studies
and a lower prevalence of Type D personality. However, there was a
nonsignificant trend in the wrong direction in terms of prediction of
being explained so that the possibility can be dismissed of the trend
likely being reversed with continued follow-up.
used to identify Type D patients in previous studies and the overfitted
regression models used in testing multivariate prediction controlling
for potential confounders . These factors would be expected to
lead to spurious findings of a prognostic value for Type D. These points
were echoed and amplified in an accompanying editorial .
Is a personality type tenable conceptually and statistically?
The Tilburg group identified patients as having a Type D personality
scored above 10 on continuous measures of both negative affectivity
(NA) and social inhibition (SI), and then compared the outcomes of
these patients in the high/high quadrant to the other three quadrants.
Such an analytic strategy for combining continuous predictor variables
has been widely rejected for decades in the personality, education,
andbasic statisticsliteratures[21–27]becauseitlikelyleads tospurious
results. There are two objections: splitting continuous variables into ar-
bitrarily dichotomized categories, but alsothe isolationofthehigh/high
category as a personality type.
Whereas dichotomizing a single continuous variable loses informa-
tion and statistical power, cross tabulation of a pair of dichotomized
variables is likely to lead to inflated estimates of statistical significance
and dramatically increased risk of spurious findings . Humphreys
 declared the construction of a 2×2 matrix from continuous vari-
ables to be “unnecessary, crude, and misleading” (p. 874) and Cohen
and Cohen called it “an abuse of data” (p. 310). In labelinga person
as an introvert or as neurotic, theorists adopted a typological language
as “a verbal convenience rather than a meaningful mode of categoriza-
tion” (p. 1159)  without any evidence that a specific demarcation
point exists, at which persons above this point not only resemble each
other, but differ from persons below it in crucial ways .
While the success of typological thinking in biology can be dem-
onstrated in the classification of plants and animals, examples in psy-
chology are harder to identify. Thus, biological sex as male or female
approaches a sharply defined category, but the representation of the
psychological dimensions of masculinity and femininity does not
. Applied to the assessment of Type D personality, the implica-
tions are for a strong preference for preserving NA and SI as continu-
ous variables and examining their interaction, rather than postulating
a sharp distinction between high/high quadrant and the other three
quadrants in a dichotomization of these variables. Among the problems
in isolating the high NA/high SI quadrant to construct Type D personal-
ed for greater distress than if either of the measures were considered
separately. Scoring high on two correlated measures of distress is a
more reliable indication of distress than being high on only one.
Long-standing concerns about statistical and conceptual difficul-
ties of categorical/typological constructs have been substantiated in
recent work usingsophisticated simulated comparisondata techniques
[29,30] that consistently favors dimensional conceptualizations. One
systematic review compared377 articleswith311 distinct findings
The domains of normal personality, mood disorders, anxiety disor-
ders, eating disorders,externalizingdisorders, andpersonalitydisor-
ders (PDs) other than schizotypal yielded little persuasive evidence
of taxa [categories]…. This review indicates that mostly variables of
interest to psychiatrists and personality and clinical psychologists
are dimensional, andthatmany influential taxonic findings of earlier
taxometric research are likely to be spurious (p. 903).
Ferguson and colleagues  applied two taxometric procedures
MAMBAC and MAXCOV to scores from the DS14, the standard measure
of Type D personalityand foundclear evidence fora representation of it
in dimensional rather categorical terms. The strong suggestion of these
tiplicative influence of NA and SI in the context of other risk factors.
While it is possible that inflection points for the continuous variables
ular CVD outcomes, the task remains of determining whether such
inflection points need to be identified for NA and SI separately or
whether there should be single or sliding cutpoints specified for the
interaction term with different outcomes.
Rescuing Type D personality as an interaction term in a regression
analysis of continuous variables?
There are multiple reasons for doubting the validity of earlier claims
from small studies that categorical Type D personality defined by the
high/high quadrant predicted mortality. The two larger studies
conducted outside the Tilburg group [17,18] had analyzed Type D per-
sonality data using both the scoringprocedure of the original investiga-
tor group, but also by preserving NA and SI as continuous variables and
examining their interaction effect, and in neither instance found a sig-
nificant bivariate or multivariate prediction of mortality. Could claims
about Type D personality's independent prognostic value nonetheless
be revived by reanalysis of the data from the earlier studies that pre-
served the continuous nature of component variables? Probably not. It
is unlikelythat a significantinteractioneffectwouldsupport categorical
over dimensional conceptions of Type D personality or that the optimal
ingofTypeD personalityin past studies.More basically, given the small
likely that the interaction term would prove significant. Smith 
pointed out that even if NA or SI alone or some additive combination
predicted mortality, there was the risk of what Block  has termed
the “jangle fallacy” of assuming that simply assigning a new name to
previously studied traits constitute a discovery.
However, Smith  provided a more profound criticism of relying
on the one cell versus three (high NA and high SI versus the three
other combinations of these variables) to test the specific prediction of
a synergistic effect of NA and SI. The three versus one contrast could
occur for numerous other reasons than a synergy of these two variables.
Namely, a significant contrast could be produced by an additive effect of
an interaction. We add that a spurious interaction effect due to either an
benefits of combining two imperfectly reliable measures of negative af-
fect rather than relying on one imperfectly reliable measure.
Meta-analysis to the rescue?
Could the entire body of studies relating Type D personality to
mortality be combined in a meta-analysis? Grande and colleagues
J.C. Coyne, J.N. de Voogd / Journal of Psychosomatic Research 73 (2012) 401–407
 performed such a meta-analysis of available studies and the
resulting forest plot (Fig. 5 in ), offered a striking graphic repre-
sentation of what was already known. The earlier smaller studies by
the Tilburg group were positive and the two later, larger scale studies
conducted outside the group are negative. With understatement, the
authors of the meta-analysis declared that “recent methodologically
sound studies suggest that early Type D studies had overestimated
the prognostic relevance (p. 299)” and ended with a call for “urgently
required,” more sophisticated studies. But these conclusions take
for granted that there is a Type D personality phenomenon to be
explained in the absence of a single positive finding having any of
the following characteristics: (a) appropriate analyses of data from
(b) an adequately powered study (c) conducted outside the original
This meta-analysis relied on the Tilburg group's categorization of a
Type D personality, rather than a regression analysis of continuous
variables and their interaction. The overall effect size was significant,
which simply attests to the ability of a substantial portion of small
underpowered studies to dominate a meta-analysis. Moreover, the
considerable heterogeneity between studies begged explanation,
without which the interpretation of the overall effect size would not
Responding to criticism  of this meta-analysis, Grande  pro-
posed a patient level meta-analysis as a solution. This would require
obtaining original continuous NA and SI data from the authors of previ-
ous studies, but for two reasons would be unsatisfactory solution and
would only serve to obscure some obvious heterogeneity and bias in
the contributing studies . First, the validity of such a patient level
meta-analysis assumes that all available data are included without sub-
D personality, even if we cannot quantify it. We had difficulty publish-
ing our large study of Type D personality predicting mortality ,
with it getting rejected from both Lancet and a number of cardiology
other cardiac outcomes. It is even less likely that negative findings from
unpowered studies would be published because of confirmatory bias
and the further justification for rejection that such studies were too
small to expect a significant finding. Second, positive studies have dif-
fered in key features such as beginning and end of observation periods,
means of assessing Type D personality, and availability and selection of
covariates. Unless we are willing to ignore such considerations, mean-
series of published underpowered studies yielding strong positive
effects involved unknown prior analyses leading to decisions about
beginning and ending of follow-up periods and selection and coding
of covariates. Think of it: would we accept the validity of a calculation
of odds based only on betting sequences that always ended with
winnings and with no losing sequences included, despite the prior
expectation that odds were favorable to the house?
Overall, obtaining complete, unbiased patient-level data for a
meta-analysis may be more difficult when the data are derived from
observational, correlational studies of the prediction of mortality
than when the data represent the declared outcomes of clinical trials.
The existence of unpublished clinical trial data may be more easily
detected than the existence of aborted examinations of mortality
data associated with observational studies, particularly when baseline
data for the study were collected for other purposes and the
follow-up assessment points were not fixed. There may be no record
of multiple examinations of data with adjustments in the start and
ending of the follow up.
Type D personality predicting mortality independent of depression
and known biomedical prognostic factors: déjà vu all over again?
Even acknowledging conceptual and statistical criticisms and the
possibility of a jangle fallacy, claims remain provocative that Type D
prognostic variables including severity of disease. These claims might
seem for some to warrant further investigation before dismissing the va-
lidity of some sort of reformulated Type D personality.
First, these claims arise in multivariate regressions with the num-
ber of covariates included in models are a substantial proportion of
the number of deaths being explained. Rules of thumb for minimal
covariates/events vary, but most authorities agree a minimum of 10
to 15 events are needed per covariates to avoid generating spurious
associations . Undoubtedly even those who advocate relaxation
of these rules would balk at having as high a ratio of covariates to
number of events being explained.
Furthermore, claims that prognostic value of Type D personality is
independent of depressive symptoms should be greeted with skepti-
cism, given the long-recognized high association among measures of
negative affect [40–44]. Seemingly reasonable theoretical and concep-
tual distinctions often cannot be sustained when attempts are made
to demonstrate discriminant validity of measures of specific negative
self-reported measures of depressive symptoms as their respective
internalconsistencies allow . Any apparentadvantage in prognostic
value of Type D personality versus depression is an artifact of the con-
struction of the high/high typology and it should be expected that
substitutingdepressive symptomsfor NAwould generate similarspuri-
ous advantages for a reformulated Type D personality. Finally, with
ther substantialcontribution topredicting mortality would be expected
from adding consideration of depressive symptoms. Whatever would
be represented by Type D personality controlling for depressive symp-
toms would have little resemblance to what is intended by the
unadjusted Type D personality variable . Statistical adjustments of
highly correlated negative affect variables can generate anomalies as
seen in a finding that anxiety is inversely related to mortality after
other negative affect variables entered into multiple regression analy-
tor when considered alone . In short, there is a large literature
relating various measures of negative affect to CVD outcomes that pro-
vides little basis for the expectation that singling out a combination of
NA and SI will have much consistent advantage over other such nega-
tive affect variables, individually or combined.
As for the claim that Type D personality predicts death independent
of established biomedical factors, it too arises in overfitted multiple
regression equations. Moreover, there is a large literature showing
that negative affect, and particularly depressive symptoms can often
befound to predictmortality, but this literature also indicates consider-
able difficulty singling out particular psychological variables as a risk
for mortality or that the association is not artifactual. Ketterer and
colleagues  have referred to the “big mush” of confounded and
non-independent negative affect measures. Meehl  has applied
the label “crud factor” to the broader tendency of self-reported neg-
ative factors to be correlated with each other in ways that cannot
readily be unambiguously differentiated. Ormel and colleagues 
have suggested that NA/neuroticism has little explanatory value be-
cause it is related to so many diverse self-reported negative anteced-
ents, concurrent conditions and consequences. He noted evidence
that NA/neuroticism involves a negative response style that extends
to other self-reports. While it might seem that a hard outcome like
mortality would escape the criticism of confounding and response
style, MacLeod and colleagues  cogently argue that the challenge
is to distinguish the causal influence of negative affect from other
negative environmental and physical health variables. They have
provided a number of demonstrations of the anomalies and inconsis-
tencies that can arise when drawing causal inferences about apparent
associations of self-reported depressive symptoms or stress with
physical health outcomes. There is a high likelihood of noncausal
J.C. Coyne, J.N. de Voogd / Journal of Psychosomatic Research 73 (2012) 401–407
relationships generated by confounding between self-reported neg-
ative affect and physical health outcomes, with residual confounding
often proving impossible to discount. What needs to be ruled out –
and generally cannot be – is that the negative affect variable merely
reflects unmeasured or imperfectly measured aspects of overall dis-
ease status, including comorbidity.
Plausible putative psychophysiological mechanisms can be invoked
to explain particular negative affects being related to mortality, but
these mechanisms often work just as well for rival negative affect vari-
ables or fora relationship found to bein theopposite direction . The
task of disambiguating any independent causal role of the negative
affect in death is shifted to the similarly difficult task of establishing
the variable's role in a poorly mapped, complex psychophysiological
process in which the precise role of easily measured physiological
variables cannot yet be identified. Showing that a particular negative
variable is correlated with some physiological variable is easier than
establishing that it mediates the causal influence of the negative
affect variable in a particular psychophysiological process .
In general, the efficacy of introducing physical health covariates has
been overestimated as a means to rule out confounding physical health
variables asthesourceof anassociationbetween self-reported negative
affect and physical health outcomes in observational studies of aged
ponent of imperfect measure of disease severity is not sufficient to dis-
count such a possibility” . This applies not only to disease severity,
but to typically overlooked physical comorbidity and overall disease
burden. A simple bivariate association of negative affect with physical
health outcomes is ambiguous and could be explained by confounding
with correlated health variables. Yet, application of statistical controls
oftenproduces a more biased and confounded estimateof the influence
of negative affecton physical healthoutcomes.The efficacy of statistical
controls depends upon the unlikely achievement of all of the relevant
potential confounders having been identified, temporal ordering
having been established, mediators having been distinguished from
confounders, and all relevant variables having been measured with
Standard practices such as introducing all covariates into regression
equations, pre-selecting them on the basis of correlations with the out-
come, or relying on automatic selection procedures such as backwards
regression all can generate considerable proportion of noise variables
, and – particularly when there is even moderate correlation
among the disease variables – exclude potentially important indepen-
dent risk factors . “Statistical adjustment by an excessive number
of variables or parameters, uninformed by substantive knowledge
(e.g. lacking coherence with biologic, clinical, epidemiological, or social
knowledge)…can obscure a true effect or create an apparent effect
when none exists.”. What is needed and has not been available in
studies of Type D personality predicting mortality is a precise model
specification. Presumed “replications” of positive findings need to
adhere to the same model or explain why another model was used,
itive results are found for Type D personality.
Does death matter anymore in psychosocial research predicting
While numerous other investigator groups have been exploring
the associations among Type D and other self-report negative affect
and health-related variables, the cachet of these studies depends on
claims from the Tilburg studies that Type D represents a unique,
strong risk factor for death. Yet, the practical clinical implications of
any association between Type D personality and mortality in CVD
may not be as clear as has been assumed. Type D personality has
been construed as a relatively immutable trait and so there would
no obvious move to intervention studies to modify Type D personali-
ty. A recent trial of a multimodal intervention failed to modify Typed
D personality . Perhaps patients with Type D personality could be
targeted for interventions that improve their disease specific health
status, but it has not been clear why it would not be more efficient
simply to target relevant physical health variables, unmoderated by
Type D personality. Calls for Type D personality being used for routine
screening of CVD patients have not been grounded in studies of the
performance characteristics of measures of Type D personality as
screening instruments and the cutpoints that have been advocated
have been arbitrary. It is not clear what the clinical application
would have been of such screening: given the small proportion of
deaths occurring in studies purporting to demonstrate a prediction
of mortality, the optimal prediction strategy would have been to dis-
regard the results of a screening with a Type D personality measure
and predict that each individual patient will survive. And given the
lack of evidence for clinically relevant predictive value, it would sim-
ply have been unethical to exclude patients from cardiac rehabilita-
tion solely on the basis of their Type D scores. The main appeal of
Type D personality seems to have been the claim that it identified
an independent predictor and therefore risk factor for mortality.
Claims that psychological variables can predict mortality and that
their modification can extend life of ill and dying people are important
to the credibility and prestige of psychosomatic medicine, and there
ables are related to mortality and even to protect these claims from dis-
confirmation. Much can be learned from the literature concerning
depression in cardiac patients. Depression is a psychological variable
that has been given particular attention in behavioral cardiology, not
only because of studies showing an association with mortality, but
because it is modifiable. Yet, a systematic review found a lack of any
studies indicatingthatroutinescreeningfordepressionimproves cardi-
ac patients' physical health outcomes . Moreover, issues have been
raised about the practical clinical relevance of mortality as an outcome.
Advances in medical treatment of cardiac conditions  now requires
as enrollment of as many as 14,000 patients to demonstrate that a clin-
ical procedure or device has an advantage over the patient survival
achievedinroutinecare. So, we are left with thedifficultiesestablishing
causality in observational studies and the likely impracticality of dem-
onstrating causality in randomized controlled trials for Type D person-
ality as well as depression.
Summary: can we save the credibility of psychosomatic medicine
from repeated decline effects?
The literature concerning Type D predicting mortality personality
seems to fit the pattern of a discovery in a seemingly underpowered
study subsequently being found to be a false positive. Not only is
there a series of larger null trials accumulating in the literature, fun-
damental issues have been raised about the validity of the conceptu-
alization of Type D personality as a categorical type. We can expect
that further studies of Type D personality will be required to analyze
personality data in terms of the interaction term of continuous vari-
ables or offer a compelling reason why the standard strategy should
continue to be used. Also, as we have seen here, further issues can
be raised when Type D personality is considered in the context of
the larger literatures relating negative affect variables to health.
Once the distinctiveness of Type D is questioned, it becomes easier
to identify difficulties progressing to an identification of Type D as a
causal factor in CVD outcomes, particularly given the precedent of
the literature concerning depression. A commentary  that accom-
panied the original Type D mortality study  in Lancet in 1996 seems
Denollet et al. added a new term – the distressed personality
(Type D) – to a field congested with related concepts including
type A personality, anger and hostility, psychological stress, vital
exhaustion, major depression, depressive symptoms, and social
J.C. Coyne, J.N. de Voogd / Journal of Psychosomatic Research 73 (2012) 401–407
isolation. Each of these concepts enjoyed a period of prime time
exposure following publication of one or more epidemiological
reports linking it to mortality in patients with CHD and then
declined in popularity…. (p. 414)
Claims that Type D personality predicted mortality attracted con-
siderable attention, even being featured in a story in Time magazine
. Such claims are newsworthy and add to the short-term credi-
bility and prestige of psychosomatic medicine. Yet, there is the risk
that a recurring pattern of a particular association declining with
the association being shown to be spurious, exaggerated, or simply
false, will undermine the credibility of the field. Previous claims
that were shown to be exaggerated or simply false contributed to
the decline and almost demise of psychosomatic medicine in the
1950s and 1960s .
It would be beneficial to reflect on the conditions that led to a series
of published underpowered studies seemingly replicating the initial
finding of Type D personality predicting mortality. Why were substan-
tially underpowered studies repeatedly accepted for publication after
the initial discovery study, rather than editors and reviewers insisting
on larger number of deaths being explained? Why were there no chal-
of evidence that dimensionswere superior to categorical conceptualiza-
tions of personality, and the data reduction analyses techniques used to
construct categories were prone to spurious results? Why was the
observation, made at the time of the first Type D personality–mortality
congested area of seemingly similar concepts ? Why was there
apparently no insistence from peer reviewers that the investigators ad-
dress these issues as a condition for publication or and why were these
criticisms not expressed in post-publication commentary like letters to
the editor? The answers to these questions may point to the proneness
of the field to confirmatory bias, particularly in claims about psycholog-
ical factors influencing mortality, and therefore the field's vulnerability
to future decline effects, as well as its limited ability to self correct in a
There is a need for reforms in the standards for evaluating claims as
“discoveries” and for publication practices in psychosomatic medicine.
True breakthrough discoveries can be expected to occur infrequently
and even when claims of discoveries are not based on confirmatory
bias and significance chasing, they can often be expected not to survive
attempts at replication. Efforts to uncover important new clinically rel-
evant phenomena are thus risky and prone to failure and so there is a
need for not makingthesearchmoredifficultthanit already is.The dis-
covery process can be expected to be “unfettered, haphazard, explor-
atory, opportunistic, selective, and highly subjectively interpreted”
[60, p 645]. Although not all would agree, some advocates for reform
suggests that flexible rules of data analysis and even fishing expeditions
are still permissible, but only if all results are reported . However, in-
vestigators should be disallowed from presenting their discoveries as if
they were obtained by conventional confirmatory, rather than explor-
atory analyses, and the full range of methods and examinations of the
data conducted before arriving at the “discovery” need to be identified
transparently. Acceptance of a discovery should only be provisional
confirmatory practices, and replication should entail obtaining precisely
similar results, rather than simply achieving statistical significance.
Across fields, there have been a number of suggestions for changes
in editorial practices that are intended to make the occurrence of de-
cline effects less common [61–65]. These include some fundamental
changes like publication of negative findings and non-replications;
prior registration of the protocols for observational studies, much
the same as clinical trials are now being registered; making publicly
available data sets for reanalysis; and not accepting “discoveries”
until they are independently replicated. A code of conduct has been
suggested for investigators that would be enforceable by reviewers
. For observational studies predicting mortality, rules might in-
clude having minimal sample sizes, evidence of pre-specified follow
up periods and analytic plans, disclosure of both simple bivariate
and multivariate results, and demonstration that positive results are
not tied to arbitrary analytic decisions.
Many of these suggestions are well intended, but run counter to
established practices that continue to be rewarded. Many journals,
and particularly high impact journals, do not publish null findings
or even replication studies, because greater prestige is accorded ini-
tial discoveries. A survey suggests that psychologists openly admit
multiple examinations of data before settling on the most positive
findings and to suppressing negative findings . A prominent psy-
chologist has openly advocated that investigators change their stated
hypotheses to what fits the results “where the data may be strong
enough to justify recentering your article around the new findings
and subordinating or even ignoring your original hypotheses.” .
Overall, publishers journals, reviewers, funding agencies, department
heads, and the media all reward extravagant claims and confirmatory
bias and outright hype .
Many of the reforms that have been suggested to reduce the pos-
sibility that most discoveries will turn out to be false or exaggerated
and a consequentinevitabledecline effect are unlikelytobe achieved
and may be unenforceable. Journals continue to resist publishing
attempted replications that fail or simply negative findings because
of the greater prestige of positive discovery studies. It also would
be difficult to enforce the requirement that investigators conducting
observational studies formally register their hypotheses, length of
follow-up, and covariates being considered before commencing
Open-minded but skeptical readers and reviewers are left to their
own devices in deciding whether new declarations of discoveries are
credible or are likely undergo a decline effect. Relying on prior prob-
abilities derived from specific research literatures but also more
general observations about the pervasiveness of decline effects, they
can ask for themselves in the case of a new discovery of a Type D
personality-like phenomenon ‘what is the likelihood of a distinct neg-
ative affect being singled out as an independent predictor of mortality
superior to its rivals and with appropriate controls for biomedical
risks factors?’ They can be skeptical of discoveries emerging in under-
powered trials that have not yet been independently replicated and
they can be skeptical of “replications” that might have been achieved
with repeated examinations of the data with flexible follow-up pe-
riods, exclusion criteria, covariates being considered and other design
and analytic decisions. In other literatures and notably in genome
wide studies, the tendency has been noted to consider particular find-
ings replicated when any positive result was achieved, even if not a
replication of the specific “discovery.” Stricter criteria for declaring
replications should be applied, if not by journal editors and reviewers,
then by skeptical readers.
 Ioannidis JPA. Contradicted and initially stronger effects in highly cited clinical
research. JAMA 2005;294:218-28.
 Ioannidis JPA. Why most published research findings are false. PLoS Med 2005;2:
 Ioannidis JPA. Evolution and translation of research findings: from bench to
where. PLoS Clin Trials 2006;1:e36.
 Schooler J. Unpublished results hide the decline effect. Nature 2011;470:437.
 Simmons JP, Nelson LD, Simonsohn U. False-positive psychology: undisclosed
flexibility in data collection and analysis allows presenting anything as significant.
Psychol Sci 2011;22:1359-66.
 Wagenmakers EJ, Wetzels R, Borsboom D, van der Maas HL. Why psychologists
must change the way they analyze their data: the case of psi: comment on Bem
(2011). J Pers Soc Psychol 2011;100:426-32.
Psychosom Med 1995;57:582-91.
independent predictor of long-term mortality in patients with coronary heart dis-
ease. Lancet 1996;347:417-21.
J.C. Coyne, J.N. de Voogd / Journal of Psychosomatic Research 73 (2012) 401–407
 Denollet J, Holmes RVF, Vrints CJ, Conraads VM. Unfavorable outcome of heart
transplantation in recipients with Type D personality. J Heart Lung Transplant
 Denollet J, Vaes J, Brutsaert DL. Inadequate response to treatment in coronary
heart disease — adverse effects of Type D personality and younger age on
5-year prognosis and quality of life. Circulation 2000;102:630-5.
 Denollet J, Pedersen SS, Vrints CJ, Conraads VM. Usefulness of Type D personality
in predicting five-year cardiac events above and beyond concurrent symptoms of
stress in patients with coronary heart disease. Am J Cardiol 2006;97:970-3.
 Pedersen SS, Denollet J, Ong ATL, Sonnenschein K, Erdman RAM, Serruys PW, et al.
Adverse clinical events in patients treated with sirolimus-eluting stents: the im-
pact of Type D personality. Eur J Cardiovasc Prev Rehabil 2007;14:135-40.
 Schiffer AA, Smith ORF, Pedersen SS, Widdershoven JW, Denollet J. Type D person-
ality and cardiac mortality in patients with chronic heart failure. Int J Cardiol
distress and mortality in systolic heart failure. Circ Heart Fail 2010;3:261-7.
 Damen N, Versteeg H, Boersma E, Serruys PW, van Geuns RJM, Denollet J, et al.
Depression is independently associated with seven year mortality in patients
treated with percutaneous coronary intervention: results from the RESEARCH
registry. Int J Cardiol 2012 (Epub ahead of print).
 Volz A, Schmid JP, Zwahlen M, Kohls S, Saner H, Barth J. Predictors of readmission
and health related quality of life in patients with chronic heart failure: a compar-
ison of different psychosocial aspects. J Behav Med 2011;34:13-22.
 Grande G, Romppel M, Vesper JM, Schubmann R, Glaesmer H, Hermann-Lingen C.
Type D personality and all-cause mortality incardiac patients — data from a German
cohort study. Psychosom Med 2011;73:548-56.
 Coyne JC, Jaarsma T, Luttik ML, van Sonderen E, van Veldhuisen DJ, Sanderman R.
Lack of prognostic value of Type D personality for mortality in a large sample of
heart failure patients. Psychosom Med 2011;73:557-62.
 Babyak MA. What you see may not be what you get: a brief, nontechnical introduc-
tion to overfitting in regression-type models. Psychosom Med 2004;66:411-21.
 Smith TW. Toward a more systematic, cumulative, and applicable science of per-
sonality and health: lessons from Type D personality. Psychosom Med 2011;73:
 Block J, Ozer JD. Two types of psychologists: remarks on the Mendelsohn, Weiss,
and Feimer contribution. J Pers Soc Psychol 1982;42:1171-81.
 Humphreys LG, Fleishman A. Pseudo-orthogonal and other analysis of variance
designs involving individual-differences variables. J Educ Psychol 1974;66:
of quantitative variables. Psychol Methods 2002;7:19-40.
Psychol Bull 1993;113:181-90.
 Royston P, Altman DG, Sauerbrei W. Dichotomizing continuous predictors in multi-
ple regression: a bad idea. Stat Med 2006;25:127-41.
 Cohen J,Cohen P. Applied multiple regression/correlation analysis for the behavioral
sciences. 2nd ed.Hillsdale, NJ: Erlbaum; 1983.
 Humphreys LG. Doing research the hard way — substituting analysis of variance
for a problem in correlational analysis. J Educ Psychol 1978;70:873-6.
 Mendelsohn GA, Weiss DS, Feimer NR. Conceptual and empirical-analysis of the
typological implications of patterns of socialization and femininity. J Pers Soc
 Haslam N, Holland E, Kuppens P. Categories versus dimensions in personality and
psychopathology: a quantitative review of taxometric research. Psychol Med
 Markon KE, Chmielewski M, Miller CJ. The reliability and validity of discrete and
continuous measures of psychopathology: a quantitative review. Psychol Bull
 Ferguson E, Williams L, O'Connor RC, Howard S, Hughes BM, Johnston DW, et al. A
taxometric analysis of Type-D personality. Psychosom Med 2009;71:981-6.
 Block J. A contrarian view of the five-factor approach to personality description.
Psychol Bull 1995;117:187-215.
 Cortina JM. Interaction, nonlinearity, and multicollinearity: implications for mul-
tiple regression. J Manag 1993;19:915-22.
 Lubinski D, Humphreys LG. Assessing spurious “moderator effects”: illustrated
substantively with the hypothesize (“synergistic”) relation between spatial and
mathematical ability. Psychol Bull 1990;107:385-93.
 Grande G, Romppel M, Barth J. Association between Type D personality and progno-
sis in patients with cardiovascular diseases: a systematic review and meta-analysis.
Ann Behav Med 2012;43:299-310.
 de Voogd JN, Sanderman R, Coyne JC. A meta-analysis of spurious associations
between Type D personality and cardiovascular disease endpoints. Ann Behav
 Grande G, Romppel M, Barth J. Type D personality and heart disease: walking the
line between enthusiasm and disbelief. Ann Behav Med 2012;44:138.
 Egger M, Schneider M, Davey Smith G. Meta-analysis — spurious precision?
Meta-analysis of observational studies. BMJ 1998;316:140-4.
 Concato J, Peduzzi P, Holfold TR, Feinstein AR. Importance of events per indepen-
dent variable in proportional hazards analysis. I. Background, goals, and general
strategy. J Clin Epidemiol 1995;48:1495-501.
 Ketterer MW, Denollet J, Goldberg AD, McCullough PA, John S, Farha AJ, et al. The
big mush: psychometric measures are confounded and non-independent in their
association with age at initial diagnosis of ischaemic coronary heart disease.
J Cardiovasc Risk 2002;9:41-8.
 Coyne JC. Self-reported distress: Analog or ersatz depression? Psychol Bull
 Ormel J, Rosmalen J, Farmer A. Neuroticism: a non-informative marker of vulner-
ability to psychopathology. Soc Psychiatry Psychiatr Epidemiol 2004;39:906-12.
 Suls J, Bunde J. Anger, anxiety, and depression as risk factors for cardiovascular
disease: the problems and implications of overlapping affective dispositions.
Psychol Bull 2005;131:260-300.
 Smith TW. Conceptualization, measurement, and analysis of negative affective
risk factors. In: Steptoe A, editor. Handbook of behavioral medicine: methods
and applications. New York, NY: Springer; 2010. p. 155-68.
 Grossardt BR, Bower JH, Geda YE, Colligan RC, Rocca WA. Pessimistic, anxious, and
depressive personality traits predict all-cause mortality: the mayo clinic cohort
study of personality and aging. Psychosom Med 2009;71:491-500.
 Meehl P. Why summaries of research on psychological theories are often
uninterpretable. Psychol Rep 1990;56:199-244.
 MacLeod J, Davey Smith G, Heslop P, Metcalfe C, Carroll D, Hart C. Psychological
stress and cardiovascular disease: empirical demonstration of bias in a prospec-
tive observational study of Scottish men. BMJ 2002;324:1247-51.
 Davey Smith G, Phillips AN, Neaton JD. Smoking as “independent” risk factor for
suicide: illustration of an artifact from observational epidemiology? Lancet
 Macleod J, Davey Smith G. Psychosocial factors and public health: a suitable case
for treatment? J Epidemiol Community Health 2003;57:565-70.
 Christenfeld NJS, Sloan RP, Carroll D, Greenland S. Risk factors, confounding, and
the illusion of statistical control. Psychosom Med 2004;66:868-75.
 Flack VF, Chang PC. Frequency of selecting noise variables in subset regression-
analysis — a simulation study. Am Stat 1987;41:84-6.
 DerksenS, KeselmanHJ.Backward, forward andstepwise automatedsubset-selection
algorithms — frequency of obtaining authentic and noise variables. Br J Math Stat
 Breslow N. Design and analysis of case–control studies. Annu Rev Public Health
 Albus C, Bjarnson-Wehrens B, Gysan DB, Herold G, Schneider CA, zu Eulenburg C,
et al. The effects of multimodal intervention for the primary prevention of cardio-
vascular diseases on depression, anxiety, and Type-D pattern. Initial results of the
randomized controlled PreFord trial. Herz 2012;37:59-62.
 Thombs BD, de Jonge P, Coyne JC, Whooley MA, Frasure-Smith N, Mitchell AJ, et al.
Depression screening and patient outcomes in cardiovascular care a systematic
review. JAMA 2008;300:2161-71.
 Lesperance F, Frasure-Smith N. The seduction of death. Psychosom Med 1999;61:
 Lesperance F, Frasure-Smith N. Negative emotions and coronary heart disease:
getting to the heart of the matter. Lancet 1996;347:414-5.
 ParkA.Are you a Type D personality?Your heart may beat risk. Time September 14,
 Holroyd KA, Coyne J. Personality and health in the 1980s — psychosomatic-
medicine revisited. J Pers 1987;55:359-75.
 Ioannidis JPA. Why most discovered true associations are inflated. Epidemiology
 Loder E, Groves T, Macauley D. Registration of observational studies. BMJ 2010;340.
 Nosek BA, Spies J, Motyl M. Scientific utopia: II. Restructuring incentives and prac-
tices to promote truth over publishability. Perspect Psychol Sci (in press).
 Schroter S, Black N, Evans S, Godlee F, Osorio L, Smith R. What errors do peer re-
viewers detect, and does training improve their ability to detect them? J R Soc Med
 John LK, Loewenstein G, Prelec D. Measuring the prevalence of questionable research
practices with incentives for truth-telling. Psychol Sci 2012;23:524-32.
 Bem DJ. Writing the empirical journal article. In: Darley JM, Zanna MP, Roediger
HL, editors. The complete academic: a practical guide for the beginning social sci-
entist. 2nd edition. Washington, DC: American Psychological Association; 2003.
 Caulfield T, Condit C. Science and the sources of hype. Public Health Genomics
J.C. Coyne, J.N. de Voogd / Journal of Psychosomatic Research 73 (2012) 401–407