Changes in Functional Health, Changes in Medication Beliefs, and
Benjamin Schu ¨z and Susanne Wurm
German Centre of Gerontology
Jochen P. Ziegelmann
Freie Universita ¨t Berlin
Lisa M. Warner
German Centre of Gerontology and Freie Universita ¨t Berlin
German Centre of Gerontology
Freie Universita ¨t Berlin
Objective: Medication adherence often lies below recommendations although it is crucial for effective
therapies, particularly in older adults with multiple illnesses. Medication beliefs are important factors for
individual adherence, but little is known about their origin. We examine whether changes in functional
health predict changes in medication beliefs, and whether such changes in beliefs predict subsequent
medication adherence. Design: At three points in time over a 6-month period, 309 older adults (65–85
years) with multiple illnesses were assessed. Latent true change modeling was used to explore changes
in functional health and medication beliefs. Adherence was regressed on changes in beliefs. Main
Outcome Measures: Medication beliefs were measured by the Beliefs About Medicines Questionnaire;
medication adherence by the Reported Adherence to Medication Scale. Results: Functional health and
medication beliefs changed over time. Increasing physical limitations predicted increases in specific
necessity and specific concern beliefs, but not in general beliefs. Changes in specific necessity beliefs
predicted intentional adherence lapses, changes in general overuse beliefs predicted unintentional
adherence lapses. Conclusions: Medication beliefs partly depend on health-related changes, and changes
in beliefs affect individual adherence, suggesting to target such beliefs in interventions and to support
older adults in interpreting health changes.
Keywords: medication adherence, medication beliefs, older adults, functional health, symptoms
Sooner or later, increasing age is almost inevitably associated
with a growing number of chronic and acute illnesses (Fried,
2000). This individual accumulation of illnesses (multimorbidity)
affects the majority of individuals over 60 years of age. Epidemi-
ological data suggest that around 61% of men and 65% of women
over 60 years suffer from two or more co-occurring diseases (van
den Akker, Buntinx, Metsemakers, Roos, & Knotterus, 1998).
Among the most common conditions in multimorbidity are
cardiovascular diseases, diabetes, and arthritis (Laux, Kuehlein,
Rosemann, & Szecsenyi, 2008; Schu ¨z, Wurm, Warner, & Tesch-
Römer, 2009). Successful treatment of these illnesses, which
means stable or improved quality of life and physical functioning,
requires, among others, the adherence to prescribed medication.
Systematic reviews have shown that poor adherence to medication
is associated with increased health problems and mortality (Simp-
son et al., 2006). Despite this, however, between 25% (DiMatteo,
2004) and 50% (World Health Organization, 2003) of adults are
not adherent to their prescribed medication.
It is important to distinguish whether medication nonadherence
is the result of accidental slips (e.g., forgetting) or a purposeful
behavioral decision. Accordingly, unintentional nonadherence and
intentional nonadherence have been differentiated (Horne, 2006;
Wroe, 2002). While unintentional nonadherence mainly means
accidental nonadherence episodes, intentional nonadherence re-
lates to changing or skipping medication on purpose. Unintentional
nonadherence is assumed to be affected by demographic and
clinical factors, whereas intentional nonadherence is assumed to be
dependent on individual beliefs about medication such as expect-
ing side effects or doubts in the necessity of medication (Wroe,
This underlines the importance of research on factors that can
shed light on individual decisions relating to medication adher-
Accepted under the editorial term of Robert M. Kaplan.
Benjamin Schu ¨z and Susanne Wurm, German Centre of Gerontology;
Jochen P. Ziegelmann, Department of Health Psychology, Freie Universita ¨t
Berlin; Lisa M. Warner, German Centre of Gerontology and Department of
Health Psychology, Freie Universita ¨t Berlin; Clemens Tesch-Römer, Ger-
man Centre of Gerontology; and Ralf Schwarzer, Department of Health
Psychology, Freie Universita ¨t Berlin.
This study (PREFER) was funded by the German Federal Ministry of
funded within this project, the third author is funded by Grant No. 01ET0801
by the same funding body. The content is the sole responsibility of the authors.
Correspondence concerning this article should be addressed to Benjamin
Schu ¨z, Ph.D., German Centre of Gerontology, Manfred-von-Richthofen-
Str. 2, 12101 Berlin, Germany. E-mail: email@example.com
2011, Vol. 30, No. 1, 31–39
© 2011 American Psychological Association
0278-6133/11/$12.00 DOI: 10.1037/a0021881
ence. While demographic and clinical factors seem to play only a
minor role in explaining who is more likely to intentionally skip
medication (DiMatteo, 2004), individual beliefs about medication
are more likely to determine individual adherence behavior
The most commonly used psychosocial approach to explore
such medication beliefs is the Necessity-Concerns Framework
(NCF; Horne, 2003), a treatment-oriented extension of the
Common-Sense Model of health and illness (CSM; Leventhal,
Brissette, & Leventhal, 2003). The theory assumes that individual
adherence behavior is an attempt to cope with the threat caused by
illnesses and results from a cognitive and emotional appraisal
process. Accordingly, people are more likely to be adherent if they
perceive their conditions as severe and threatening enough to
warrant treatment, and at the same time perceive a specific treat-
ment to have more positive than negative consequences. For ex-
ample, a person with cardiovascular disease who experiences
worsening in symptoms, such as increasing difficulties to climb
stairs, and who perceives these symptoms to be serious enough to
treat them, might come to the conclusion that adhering to the
prescribed cardiovascular medication promises more positive (e.g.,
symptom relief) than negative (e.g., drowsiness) results and thus
take the medication more regularly than before.
These beliefs can be measured with the Beliefs about Medicines
Questionnaire (BMQ; Horne, Weinman, & Hankins, 1999), that
assesses cognitions both specific to an individual’s prescriptions
(specific necessity: individual beliefs in the necessity and benefi-
cial effects of medication; specific concerns: individual beliefs in
negative consequences of medication such as side effects) and
general for all medicines (general harm: beliefs that medicines do
more harm than good; general overuse: beliefs that too many
medicines are being prescribed). The BMQ has been applied in a
multitude of clinical settings from medication for cardiovascular
disease (Bane, Hughes, & McElnay, 2006) and arthritis (Neame &
Hammond, 2005), or diabetes (Tibaldi et al., 2009). A result
consistent in most illness groups was that specific beliefs were
better predictors of adherence than general beliefs.
Medication Beliefs in Older Adults With Multiple
Older adults with multiple illnesses have specific need to adhere
to their prescribed medication, yet adherence falls below desirable
levels. Our study therefore aims at exploring whether medication
beliefs predict medication adherence in this specific sample. Indi-
vidual beliefs about medication might be especially important
factors for adherence in older people with multiple illnesses: The
likelihood of potentially inappropriate medication, drug interac-
tions and application problems increases with the number of med-
icines prescribed (Milton, Hill-Smith, & Jackson, 2008). This can
affect how older people think about their medication (Modig,
Kristensson, Kristensson Ekwall, Rahm Hallberg, & Midlov,
2009), which in turn changes individual adherence (Benner et al.,
2009; Hughes, 2004). One challenge in examining older individ-
uals with multiple illnesses using the CSM and NCF is that
individuals who suffer from multiple illnesses experience both
symptoms that are specific to one disease and symptoms that can
be caused by several diseases, which complicates assessing spe-
cific attributions of symptoms to illnesses and treatments.
The illnesses most prevalent in multimorbid older adults are
cardiovascular disease, diabetes or arthritis (Laux et al., 2008;
Schu ¨z et al., 2009). These illnesses have in common that they are
often accompanied by functional limitations (Kadam & Croft,
2007; Noe ¨l et al., 2007). The majority of individuals affected by
multiple illnesses experience functional limitations as highly prev-
alent symptoms during their illness trajectories, as many illnesses
share risk factors and symptoms (Laux et al., 2008). Thus, our
study, rather than focusing on a list of symptoms as often em-
ployed in studies examining the CSM and the NCF, examines
functional limitations because they reflect shared symptoms of
The CSM and NCF assume that on experiencing an increase in
illness-related symptoms, individuals enter a contemplation pro-
cess, which finally results in taking up adaptive behavioral re-
sponses (in this case, medication adherence) or not. There is
evidence for this process: Halm, Mora, and Leventhal (2006) have
shown that a substantial proportion of older adults only consider
themselves as suffering from asthma and adhere to treatment when
experiencing concrete symptoms, whereas adherence was poor
during asymptomatic episodes. For the target group of multimor-
bid older adults, it is therefore crucial to examine whether changes
in physical functioning lead to changes in cognitions, which in turn
predict changes in behavior. Such change-change associations are
inherent in most behavior-oriented theories, yet have rarely been
examined empirically (Weinstein & Rothman, 2005).
Our study aims at examining the associations between changes
in health, changes in medication beliefs, and changes in medica-
tion adherence implied in the Common-Sense Model (Leventhal,
Brissette, & Leventhal, 2003) and the Necessity-Concerns-
Framework (Horne, 2003) in a sample of older adults suffering
from multiple illnesses, a group at specifically high risk for further
health deteriorations. To our knowledge, these change-change
associations have not been tested before. We rely on latent true
change modeling (Geiser, Eid, Nussbeck, Courvoisier, & Cole,
2010; Steyer, Partchev, & Shanahan, 2000) to model individual
change-change associations using a structural equation modeling
framework. In particular, we will examine whether changes in
physical functioning as a symptom prevalent in most multimor-
bidity patterns can predict changes in individual beliefs about
medication. We will further examine the shape of change, that is,
whether decreases or increases in physical functioning predict
decreases or increases in medication beliefs. Finally, we will
examine whether such changes in medication beliefs can predict
subsequent medication adherence.
These research questions were addressed in a longitudinal study
with three measurement points in time over a 6-month period.
Participants and Procedure
Participants for this study were recruited from the third assess-
ment wave of the German Aging Survey (Wurm, Tomasik, &
Tesch-Römer, 2010), a population-based representative survey in
SCHU¨Z ET AL.
the German population between 40 and 85 years of age. Partici-
pants were considered eligible if they fulfilled the following in-
clusion criteria: a) being 65 years of age or older; b) suffering from
at least two conditions mentioned either in the Charlson Comor-
bidity Index (Charlson, Szatrowski, Peterson, & Gold, 1994) or the
Functional Comorbidity Index (Groll, To, Bombardier, & Wright,
2005); and c) having provided written consent to be contacted for
further studies. From the initial sample of 6,205 people (aged
40–85), n ? 443 (7.14%) fulfilled all inclusion criteria. Of these
443, n ? 309 (69.7%) gave informed consent for the study.
Participants in this baseline sample were on average 73.27 years
old (SD ? 5.1), and 41.7% of them were women. Around 12.6%
indicated low, 52.1% medium, and 35.3% high education accord-
ing to the International Standard Classification of Education
(Unesco, 1997). Geographically, participants came from all re-
gions of Germany, with n ? 108 (35%) living in the Eastern
federal states (former German Democratic Republic).
After giving informed consent, participants made appointments
for the first measurement (Time 1, March 2009). They were then
visited at home by trained interviewers, completed a 30-min in-
terview and were given a questionnaire with a prepaid return
envelope. The second measurement point (Time 2, June 2009) was
a questionnaire only with prepaid return envelope. Here, n ? 252
(81.6% of the initial sample) took part. The third measurement
point (Time 3, September 2009) consisted of another personal
interview and questionnaire. This assessment was completed by
n ? 271 participants (87.7% of the initial sample). At every
measurement point, participants not returning the questionnaire
received one postal reminder.
A full medication inventory (Psaty et al., 1992) was conducted
at Time 1: Participants were asked to bring all medicines which
were then recorded according to the medication code (if available)
or the drug brand name and dosage. Beliefs about Medicines were
assessed at Times 1, 2, and 3 using a 8-item short version of the
Beliefs about Medicines Questionnaire (Horne et al., 1999). This
short version was developed from data of a pilot study in N ? 104
older individuals with multiple illnesses by reducing the original
questionnaire to the best-discriminating items for the four BMQ
subscales (two items each): General Harm (e.g., “Most medicines
are addictive,” interitem correlation rii? .44, p ? .01); General
Overuse (e.g., “Doctors use too many medicines,” rii? .64, p ?
.01); Specific Necessity (e.g., “My health, at present, depends on
my medicines,” rii? .79, p ? .01); and Specific Concerns (e.g.,
“My medicines disrupt my life,” rii? .56, p ? .01). All items were
answered on 5-point Likert-type scales ranging from 1 “totally
disagree” to 5 “totally agree.” Medication Adherence was mea-
sured by two items from the Reported Adherence to Medication
scale (RAM; Horne et al., 1999) asking about altering and forget-
ting to take medication.
Physical functioning was measured in the questionnaire by
using the 10-item Physical Functioning subscale of the SF-36
(Ware & Sherbourne, 1992). The degree of limitation in activities
such as lifting or carrying groceries, kneeling, walking, bathing,
dressing, and so forth was rated on a three-point scale from 1
“severely limited” to 3 “not limited at all.” In order to gain more
reliable estimates, the 10 items were parceled into two parcels of
five items each (Bandalos, 2002). Illnesses were assessed by
asking participants to indicate their illnesses on a list of 23 con-
ditions informed by the Charlson Comorbidity Index (Charlson,
Szatrowski, Peterson, & Gold, 1994) and the Functional Comor-
bidity Index (Groll et al., 2005). Covariates included age, sex,
number of illnesses, and educational status according to the Inter-
national Standard Classification of Education (Unesco, 1997).
Changes in physical functioning and changes in medication
beliefs were analyzed using Latent True Change Modeling (LTC;
Geiser et al., 2010; Reuter et al., 2010; Steyer, Eid, & Schwenk-
mezger, 1997; Steyer et al., 2000). Separate models were com-
puted for the four medication beliefs domains specific necessity
(Model 1), specific concerns (Model 2), general harm (Model 3),
and general overuse (Model 4). LTC factor scores were saved, and
adherence (log-transformed RAM scores due to severe skewness)
was subsequently regressed on the LTC factors. Analyses of vari-
ance were computed in order to illustrate the shape of change in
physical functioning and medication beliefs. Dropout analyses
were conducted to examine differences between participants drop-
ping out of the study and those remaining in the study. All latent
analyses were done in MPlus 5 (Muthe ´n & Muthe ´n, 2007), with
missing data specified in the model using the Maximum Likeli-
hood Estimator with robust standard errors. All other analyses
were done in SPSS. Goodness of fit was evaluated using the
?2-test, the Comparative Fit Index (CFI) and the Root Mean
Square Error of Approximation (RMSEA). Good model fit is
indicated by a nonsignificant ?2-test, a CFI ? .95 and a low
RMSEA value (p ? .05).
Latent True Change Modeling.
eling (LTC) is an approach to modeling individual differences in
intraindividual change based on structural equation modeling with
all advantages of latent variables, in particular greater reliability
due to the control of measurement error. Compared to growth
curve modeling, LTC has the advantage of allowing for individual
differences in change. In principle, LTC models are an extension
of the basic longitudinal measurement model of confirmatory
factor analysis that additionally decomposes the time-specific fac-
tors of subsequent measurement points into the state factor of the
previous measurement point plus a latent change factor represent-
ing latent change between the measurement occasions. (Geiser et
al., 2010; Reuter et al., 2010). This allows for identifying and
measuring the latent change factor using the indicators for the state
factor and relating this latent change factor to other variables
within a model.
In addition, indicator-specific factors were specified in order to
control for indicator-specific effects due to for example, unique
item wordings, that longitudinally can amount to a serious problem
in model specification, as this would lead to higher autocorrela-
tions of a specific indicator over time than the correlations with
other indicators measuring the same construct (Eid, Schneider, &
Schwenkmezger, 1999; Reuter et al., 2010).
The generic LTC model in Figure 1 illustrates that changes in
physical functioning between Time 1 and Time 2 predict changes
in medication beliefs between Time 2 and Time 3 (?123). In
addition, changes in physical functioning between Time 1 and
Time 2 were also specified as predictors of concurrent changes in
Latent True Change Mod-
CHANGES IN MEDICATION BELIEFS
medication beliefs (?122), and changes in physical functioning
between Time 2 and Time 3 predicted concurrent changes in
medication beliefs (?133).
Participants suffered on average from 5.55 illnesses (SD ?
2.99), with hypertension (67.64%), ostheoarthritis (63.11%), hy-
perlipidemia (49.19%), arthritis (31.07%), and peripheral vascular
disease (30.74%) being the five most prominent conditions. On
average, participants consumed 4.26 (SD ? 2.96) medicines per
Participants dropping out between Time 1 and Time 2 or be-
tween Time 2 and Time 3 were examined for significant differ-
ences in the study variables at Time 1. Drop-outs at Time 2
indicated significantly higher specific necessity beliefs and signif-
icantly lower general overuse beliefs at Time 1 (all ps ? .05). No
significant differences were found for those who dropped out
between Time 1 and Time 3.
Goodness of Fit
Model 1 (LTC model for changes in specific necessity) fitted the
data well with CFI ? .99 and RMSEA ? .03, however, the ?2-test
was significant: ?2? 99.51, df ? 77, p ? .05. Model 2 (changes
in specific concerns) showed a similarly good fit with CFI ? .99,
RMSEA ? .03 and ?2? 97.14, df ? 77, ns. Fit for general
medication beliefs was less promising: Model 3 (changes in
general overuse beliefs) yielded CFI ? .99, RMSEA ? .03, but
?2? 101.83, df ? 77, p ? .05. Model 4 (changes in general
harm) yielded CFI ? .95, RMSEA ? .08 and ?2? 256.80, df ?
77, p ? .01.1
Latent Change Models
All means of the latent change variables were significantly
different from zero, which indicates substantial change in the
factors over the three measurement points in total and between
neighboring measurement points.
Table 1 shows that changes in specific necessity beliefs (Model
1) between Time 2 and Time 3 were significantly predicted by
changes in physical functioning between Time 1 and Time 2
(?123? ?.25, p ? .01) in the manner that increases in physical
functioning predicted decreases in specific needs. Similarly,
1Detailed factor loading and variance component tables are available
from the first author upon request.
2In order to identify the model, the first indicator of each factor is
specified as reference indicator and is accordingly fixed at one. Note that
the T2 and T3 factors are perfectly determined by the T1 state factors and
the T1–T2 change factors (for the T2 factors) and in addition the T2–T3
change factors (for the T3 factors). The factor loadings ? have no occasion
index k, because factor loadings are time-invariant.
Time 1, 2, and 3.2
Generic Latent True Change Model for Physical Functioning (PF) and Medication Beliefs (MB) at
SCHU¨Z ET AL.
changes in specific concerns (Model 2) between Time 2 and Time
3 were significantly predicted by changes in physical functioning
between Time 2 and Time 3 (?133? ?.21, p ? .01). Changes in
general medication beliefs (general harm, Model 3 and general
overuse, Model 4) on the other hand were not significantly pre-
dicted by changes in physical functioning.
Analyses of variance were conducted using quartiles of LTC
scores in physical functioning between Time 1 and Time 2 as
factor, and LTC scores in medication beliefs between Time 2 and
Time 3 as dependent variables in order to illustrate these relations.
Both ANOVAs for changes in specific necessity, F(3, 303) ?
5.84, p ? .01 (linear term) and changes in specific concerns, F(1,
303) ? 2.89, p ? .01 (quadratic term) were significant. Figure 2
shows that in particular individuals in the lower quartile of phys-
ical functioning change (i.e., those with decreases) perceived in-
creases in specific necessity of their medication. Specific concern
beliefs decreased strongest in individuals in the fourth quartile of
physical functioning change (improvements).
Covariates of Medication Beliefs
Table 2 shows that the health-related covariates (number of
illnesses and number of medicines) affected all medication beliefs
in that more medicines were associated with higher specific ne-
cessity and specific concern, but with less general harm and
general overuse. A higher number of illnesses were associated with
higher specific concern, higher general harm and higher general
overuse. From the socioeconomic covariates, only education pre-
dicted medication beliefs in that less education was associated with
higher specific concerns and higher general harm beliefs.
Changes in Medication Beliefs and Changes in
In a third set of analyses, self-reported intentional and uninten-
tional nonadherence (log-transformed due to skewness) were re-
gressed on the LTC scores (see Table 3). The hierarchical multiple
regression analyses showed that intentional nonadherence at Time
3 could be predicted by baseline behavior (nonadherence) in the
first step and the number of medications in the second step, with
more medicines making intentional nonadherence less likely. In
the third step, changes in general harm and specific necessity
predicted nonadherence, with increasing general harm beliefs mak-
ing nonadherence more likely, and increasing specific necessity
beliefs making nonadherence less likely. For unintentional nonad-
herence Time 3, baseline behavior was predictive in the first step,
LTC Scores of Specific Necessity and Specific Concern in Quartiles of LTC Scores Physical
Parameter Estimates for Latent True Change Models 1–4
?p ? .05.
??p ? .01.
CHANGES IN MEDICATION BELIEFS
and increases in general overuse perceptions were predictive of
increases in the likelihood to forget to take medication.
This study examined whether changes in subjective functional
health caused changes in medication beliefs in a sample of older
adults suffering from multiple illnesses using a latent true change
modeling approach, and whether such changes in beliefs could
predict self-reported medication adherence behavior.
The operationalisation of medication beliefs in this article relies
on the Beliefs about Medicines Questionnaire (BMQ; Horne et al.,
1999), which assumes that individuals hold both specific necessity
and concern beliefs about their own medication and general beliefs
about medication. We found that changes in functional health
predicted changes in specific medication beliefs of the participants,
but not changes in general beliefs about medicines. Changes in
specific necessity beliefs predicted intentional adherence behavior,
whereas changes in general overuse beliefs predicted unintentional
Changes in Functional Health and Changes in
In particular, we found that decreases in functional health pre-
dicted subsequent increases in specific necessity beliefs and con-
current increases in specific concern beliefs.
Participants with decreasing functional health perceived increas-
ing necessity for their medication, whereas participants with in-
creasing functional health perceived decreasing necessity for their
medication. These findings can be interpreted in line with the
Necessity-Concerns Framework for medication adherence (NCF;
Horne, 2003), an extension of the treatment control facet of the
Common-Sense Model of health and illness behavior (CSM; Lev-
enthal et al., 2003). In terms of the CSM, an increase in symptoms
of the illness can lead to adaptive responses such as higher per-
ceived necessity of medication. This means that the perceived
necessity for medication due to increased symptoms outweighs
possible concerns about the medication (Clifford, Barber, &
Horne, 2008; Horne & Weinman, 2002). Our study adds to pre-
vious work in that it establishes a longitudinal perspective on the
genesis of such beliefs in treatment necessity, which allows for
stronger causal interpretations. The finding that, on the other hand,
in participants with improving functional health, specific necessity
beliefs decreased, might seem counterintuitive at first sight—after
all, symptoms have improved, which could also be seen as an
indicator of the effectiveness of medication. However, previous
work in asthma patients has shown that the experience of illness
symptoms is central to perceiving a necessity of treatment for
chronic illness (Halm et al., 2006). In addition, this finding can be
interpreted in line with Horne’s (2003) observation that most
individuals taking medication seem to hold negative medication
schemes: As soon as symptoms alleviate, the specific need to take
Parameter Estimates for Medication Beliefs Time 1 and Physical Functioning Time 1 Regressed on Covariates in Models 1–4
Model 1 Model 2Model 3 Model 4
SN T1PF T1 SC T1 PF T1GH T1 PF T1GO T1 PF T1
Sex (1 ? Male, 2 ? Female)
Number of illnesses
Number of medicines
?p ? .05.
SN ? specific necessity; SC ? specific concerns; GH ? general harm; GO ? general overuse; PF ? physical functioning.
??p ? .01.
Intentional and Unintentional Nonadherence to Medication Time 3 Regressed on LTC Scores and Covariates
Intentional nonadherence time 3Unintentional nonadherence time 3
? Step 1
? Step 2
? Step 3
? Step 1
? Step 2
? Step 3
Nonadherence Time 1
Functional health time 1
LTC general harm
LTC general overuse
LTC specific necessity
LTC specific concern
?p ? .05.
??p ? .01.
SCHU¨Z ET AL.
further medication decreases, and potentially negative general
medication schemes overwrite cognitive representations of posi-
tive effects of medication adherence as expressed in the necessity
We found a synchronic effect of changes in functional health on
changes in specific concerns about medication. In particular, de-
creasing functional health was accompanied by increases in con-
cerns about medication, and increasing functional health was ac-
companied by decreases in medication concerns. This is consistent
with the few previous studies examining the effects of symptom
change on medication beliefs (Aikens, Nease, & Klinkman, 2008;
Gonzalez et al., 2007; Schrimshaw, Siegel, & Lekas, 2005), which
suggest that participants who have experienced positive effects of
treatment decrease possible concerns such as expecting side ef-
fects. Our finding that changes in functional health concurrently
predict changes in medication beliefs gives further support to the
idea that individuals do take into account their specific illness-
related experiences when evaluating possible concerns about their
medication, which is in contrast to Horne and Weinman’s (2002)
hypothesis that such concerns are mainly fueled by general beliefs
The finding that general beliefs about medicines were not af-
fected by changes in functional health is in accordance with the
CSM, as such general beliefs are thought to rely on general
schemes about medicines, that in turn are fueled by media, previ-
ous experience, and general critical views toward science and
medicine (Horne, 2003).
In addition to these change-change associations, we found that
a higher number of illnesses was associated with more specific
concerns, more general harm beliefs and more general overuse
beliefs. This suggests that individuals with worse health are more
concerned about their medicines. Previous research suggests that
such associations might be due to an unclear attribution of symp-
toms and impairments to either illnesses or treatment side effects:
Individuals could thus attribute symptoms to their medication and,
in turn, gain more negative beliefs about their medication (Schrim-
shaw, Siegel, & Lekas, 2005). The number of medicines an indi-
vidual was taking was negatively related to general harm and
general overuse beliefs, but positively related to specific necessity
and specific concerns. This could, on the one hand, suggest that
individuals who take more medicines are in fact in greater need,
but also that they might worry more about possible side effects of
their medication regimen (Modig et al., 2009). We also found that
education was negatively related to specific concerns and general
harm beliefs, which suggests that particularly individuals with
lower education are at risk for negative treatment beliefs. Similar
results have been reported by Mårdby, Åkerlind, and Jörgensen
(2007), who found education to negatively affect medication be-
Changes in Medication Beliefs and Changes in
Apart from examining changes in medication beliefs, our study
goes beyond previous studies on the relation between medication
beliefs and adherence in that its longitudinal latent true change
design supports a causal chain from changes in illness symptoms
(functional limitations) over changes in medication beliefs (in
particular specific necessity and specific concerns) to adherence
creases in specific necessity beliefs predicted better adherence to
medication in that individuals reported less intentional changes in
their adherence behavior. This corroborates a large body of re-
search on medication beliefs that consistently shows that necessity
beliefs are associated with better adherence (Clifford et al., 2008;
Horne, 2003). The change-change associations implied in the NCF
have however not yet been examined.
Individuals with improving health, however, perceive less spe-
cific necessity and, thus, are at greater risk for nonadherence
(Horne & Weinman, 2002). In addition to that, increases in the
general belief that medication is harmful predicted a greater like-
lihood to intentionally change or skip prescribed medication reg-
imens. Although the BMQ suggests specific medication beliefs to
be better predictors of adherence than general beliefs, some pre-
vious studies (e.g., Mårdby et al., 2007) reported similar effects of
general medication beliefs on adherence.
adherence problems, we found that in particular individuals who
increased their beliefs in physicians prescribing too many medi-
cines were more likely to forget to take their medication. This
finding suggests that there might be motivated memory effects at
work, in that people are more likely to forget medication adherence
if they value medication negatively.(congeniality hypothesis; Ea-
gly, Chen, Chaiken, & Shaw-Barnes, 1999).
In particular, we found that in-
With regard to unintentional
Functional Health and Symptoms in Multimorbidity
Our study relies on the CSM (Leventhal, Brissette, & Leventhal,
2003) as theoretical background. Experiencing symptoms is the
central input in the process depicted in the CSM, and studies
usually assess this part by presenting lists of symptoms and asking
individuals to indicate whether these symptoms were caused by
their illness or not (e.g., Kaptein et al., 2007). The specific focus
of our study on multimorbid older adults, however, makes exam-
ining such attributions to specific illnesses difficult. All partici-
pants in our study suffered from two or more illnesses, at least in
part sharing risk factors and symptoms (Laux, Kuehlein, Rose-
mann, & Szecsenyi, 2008). Thus, we relied on the physical func-
tioning subscale of the SF-36 (Ware & Sherbourne, 1992) as a
well-evidenced indicator for functional limitations, which are ex-
perienced by a majority of older individuals suffering from mul-
tiple illnesses (Kadam & Croft, 2007).
Limitations and Strengths
There are limitations to our study that merit discussion. First,
adherence behavior was assessed via self-report. Socially desirable
behaviors such as medication adherence are likely to be overre-
ported (Stafford, Jackson, & Berk, 2008), and the positively
skewed distribution of self-reported adherence scores in our study
seems to confirm this hypothesis. However, comparative studies
have shown that self-reports are, nevertheless, valid indicators of
individual differences in adherence (Hansen et al., 2009). Further-
more, the sample of our study consisted of unpaid volunteers,
which makes it likely that our sample overrepresents relatively
healthy individuals with multiple illnesses. Nevertheless, the Ger-
CHANGES IN MEDICATION BELIEFS
man Aging Survey, from which our sample was selected, is a
population-representative study (Wurm et al., 2010), which im-
plies that our sample goes beyond mere convenience samples.
The operationalization of symptoms as changes in functional
health lacks the specific attribution of symptoms to illnesses as in
the IPQ, but it allows to assess changes in symptoms shared by a
multitude of illnesses in older individuals and, thus, model a chain
of changes in health, cognitions and behavior that could be gen-
eralized to the population of older individuals, who in their ma-
jority are affected by multiple illnesses rather than single, isolated
illnesses (van den Akker, Buntinx, Metsemakers, Roos, & Knot-
Apart from these limitations, our study has some specific
strengths. First, it focuses on older adults with multiple illnesses,
an underresearched high-risk group with high need for medication
adherence (Noe ¨l et al., 2007). A second strength is the use of latent
true change methodology, which allows to relate true
measurement-error-free changes in functional health to changes in
medication beliefs and, thus, provides stronger support for the
assumed causal chain between health, cognitions, and behavior.
Our study showed for the first time that changes in functional
health predict changes in medication beliefs, and that these
changes in medication beliefs can predict medication adherence in
multimorbid older adults. In particular the finding that improve-
ments in functional health lead to decreases in specific necessity
beliefs and that such necessity beliefs are important predictors of
intentional adherence suggests targeting such beliefs in interven-
tions. Such interventions might stress that even relatively few or
improving symptoms warrant medication adherence. This might be
complicated by the representations of symptomless illnesses
(Meyer, Leventhal, & Gutmann, 1985), but interpreting receding
symptoms as effects of the medication has been shown to improve
adherence (Gonzalez et al., 2007). In particular for the target group
of older adults with multiple illnesses, such interventions fostering
adaptive medication beliefs are promising for improving adher-
Aikens, J. E., Nease, D. E., Jr., & Klinkman, M. S. (2008). Explaining
patients’ beliefs about the necessity and harmfulness of antidepressants.
Annals of Family Medicine, 6, 23–29.
Bandalos, D. L. (2002). The effects of item parceling on goodness-of-fit
and parameter estimate bias in structural equation modeling. Structural
Equation Modeling, 9, 78–102.
Bane, C., Hughes, C. M., & McElnay, J. C. (2006). The impact of
depressive symptoms and psychosocial factors on medication adherence
in cardiovascular disease. Patient Education and Counseling, 60, 187–
Benner, J. S., Chapman, R. H., Petrilla, A. A., Tang, S. S., Rosenberg, N.,
& Schwartz, J. S. (2009). Association between prescription burden and
medication adherence in patients initiating antihypertensive and lipid-
lowering therapy. American Journal of Health-System Pharmacy, 66,
Charlson, M. E., Szatrowski, T. P., Peterson, J., & Gold, J. (1994).
Validation of a combined comorbidity index. Journal of Clinical Epi-
demiology, 47, 1245–1251.
Clifford, S., Barber, N., & Horne, R. (2008). Understanding different
beliefs held by adherers, unintentional nonadherers, and intentional
nonadherers: Application of the Necessity-Concerns Framework. Jour-
nal of Psychosomatic Research, 64, 41–46.
DiMatteo, M. R. (2004). Variations in patients’ adherence to medical
recommendations: A quantitative review of 50 years of research. Med-
ical Care, 42, 200–209.
Eagly, A. H., Chen, S., Chaiken, S., & Shaw-Barnes, K. (1999). The impact
of attitudes on memory: An affair to remember. Psychological Bulletin,
Eid, M., Schneider, C., & Schwenkmezger, P. (1999). Do you feel better or
worse? The validity of perceived deviations of mood states from mood
traits. European Journal of Personality, 13, 283–306.
Fried, L. P. (2000). Epidemiology of aging. Epidemiologic Reviews, 22,
Geiser, C., Eid, M., Nussbeck, F. W., Courvoisier, D. S., & Cole, D. A.
(2010). Analyzing true change in longitudinal multitrait-multimethod
studies: Application of a multimethod change model to depression and
anxiety in children. Developmental Psychology, 46, 29–45.
Gonzalez, J. S., Penedo, F. J., Llabre, M. M., Dura ´n, R. E., Antoni, M. H.,
Schneiderman, N., & Horne, R. (2007). Physical symptoms, beliefs
about medications, negative mood, and long-term HIV medication ad-
herence. Annals of Behavioral Medicine, 34, 46–55.
Groll, D. L., To, T., Bombardier, C., & Wright, J. G. (2005). The devel-
opment of a comorbidity index with physical function as the outcome.
Journal of Clinical Epidemiology, 58, 595–602.
Halm, E. A., Mora, P., & Leventhal, H. (2006). No symptoms, no asthma:
The acute episodic disease belief is associated with poor self-
management among inner-city adults with persistent asthma. Chest, 129,
Hansen, R. A., Kim, M. M., Song, L., Tu, W., Wu, J., & Murray, M. D.
(2009). Comparison of methods to assess medication adherence and
classify nonadherence. The Annals of Pharmacotherapy, 43, 413–422.
Horne, R. (2003). Treatment perceptions and self-regulation. In L. D.
Cameron & H. Leventhal (Eds.), The self-regulation of health and illness
behaviour (pp. 138–153). New York: Routledge.
Horne, R. (2006). Compliance, adherence, and concordance: Implications
for asthma treatment. Chest 130, 65S–72S.
Horne, R., & Weinman, J. (2002). Self-regulation and self-management in
asthma: Exploring the role of illness perceptions and treatment beliefs in
explaining non-adherence to preventer medication. Psychology &
Health, 17, 17–32.
Horne, R., Weinman, J., & Hankins, M. (1999). The Beliefs About Med-
icines Questionnaire: The development and evaluation of a new method
for assessing the cognitive representation of medication. Psychology &
Health, 14, 1–24.
Hughes, C. M. (2004). Medication non-adherence in the elderly: How big
is the problem? Drugs & Aging, 21, 793–811.
Kadam, U. T., & Croft, P. R. (2007). Clinical multimorbidity and physical
function in older adults: A record and health status linkage study in
general practice. Family Practice, 24, 412–419.
Kaptein, A. A., van Korlaar, I. M., Cameron, L. D., Vossen, C. Y., van der
Meer, F. J. M., & Rosendaal, F. R. (2007). Using the common-sense
model to predict risk perception and disease-related worry in individuals
at increased risk for venous thrombosis. Health Psychology, 26, 807–
Laux, G., Kuehlein, T., Rosemann, T., & Szecsenyi, J. (2008). Co- and
multimorbidity patterns in primary care based on episodes of care:
Results from the German CONTENT project. BMC Health Services
Research, 8, 14.
Leventhal, H., Brissette, I., & Leventhal, E. A. (2003). The common-sense
model of self-regulation of health and illness. In L. D. Cameron & H.
Leventhal (Eds.), The self-regulation of health and illness behaviour (pp.
42–65). Howard: Routledge.
Mårdby, A.-C., Åkerlind, I., & Jörgensen, T. (2007). Beliefs about medi-
SCHU¨Z ET AL.
cines and self-reported adherence among pharmacy clients. Patient Download full-text
Education and Counseling, 69, 158–164.
Meyer, D., Leventhal, H., & Gutmann, M. (1985). Common-sense models
of illness: The example of hypertension. Health Psychology, 4, 115–135.
Milton, J. C., Hill-Smith, I., & Jackson, S. H. D. (2008). Prescribing for
older people. British Medical Journal (Clinical Research Ed.), 336,
Modig, S., Kristensson, J., Kristensson Ekwall, A., Rahm Hallberg, I., &
Midlov, P. (2009). Frail elderly patients in primary care-their medication
knowledge and beliefs about prescribed medicines. European Journal of
Clinical Pharmacology, 65, 151–155.
Muthe ´n, L. K., & Muthe ´n, B. O. (2007). MPlus User’s Guide (5th ed.). Los
Angeles: Muthe ´n & Muthe ´n.
Neame, R., & Hammond, A. (2005). Beliefs about medications: A ques-
tionnaire survey of people with rheumatoid arthritis. Rheumatology
(Oxford, England), 44, 762–767.
Noe ¨l, P. H., Parchman, M. L., Williams, J. W., Jr., Cornell, J. E., Shuko,
L., Zeber, J. E., . . . Pugh, J. A. (2007). The challenges of multimorbidity
from the patient perspective. Journal Of General Internal Medicine, 22,
Psaty, B. M., Lee, M., Savage, P. J., Rutan, G. H., German, P. S., & Lyles,
M. (1992). Assessing the use of medications in the elderly: Methods and
initial experience in the Cardiovascular Health Study. The Cardiovas-
cular Health Study Collaborative Research Group. Journal of Clinical
Epidemiology, 45, 683–692.
Reuter, T., Ziegelmann, J. P., Wiedemann, A., Geiser, C., Lippke, S., &
Schu ¨z, B. (2010). Changes in intentions, planning, and self-efficacy
predict changes in behaviours: An application of latent true change
modeling. Journal of Health Psychology, 15, 935–947.
Schrimshaw, E. W., Siegel, K., & Lekas, H.-M. (2005). Changes in
attitudes toward antiviral medication: A comparison of women living
with HIV/AIDS in the pre-HAART and HAART eras. AIDS and Be-
havior, 9, 267–279.
Schu ¨z, B., Wurm, S., Warner, L. M., & Tesch-Römer, C. (2009). Health
and Subjective Well-Being in Later Adulthood: Different Health States,
Different Needs? Applied Psychology: Health and Well-Being, 1, 23–
Simpson, S. H., Eurich, D. T., Majumdar, S. R., Padwal, R. S., Tsuyuki,
R. T., Varney, J., & Johnson, J. A. (2006). A meta-analysis of the
association between adherence to drug therapy and mortality. British
Medical Journal: British Medical Journal, 333, 15.
Stafford, L., Jackson, H. J., & Berk, M. (2008). Illness beliefs about heart
disease and adherence to secondary prevention regimens. Psychosomatic
Medicine, 70, 942–948.
Steyer, R., Eid, M., & Schwenkmezger, P. (1997). Modeling true intrain-
dividual change: True change as a latent variable. Methods of Psycho-
logical Research - Online, 2, 21–33.
Steyer, R., Partchev, I., & Shanahan, M. (2000). Modeling true intra-
individual change in structural equationmodels: The case of poverty and
children’s psychosocial adjustment. In T. D. Little, K. U. Schnabel, & J.
Baumert (Eds.), Modeling longitudinal and multiple-group data: Prac-
tical issues, applied approaches, and specific examples (pp. 109–126).
Hillsdale, NJ: Erlbaum.
Tibaldi, G., Clatworthy, J., Torchio, E., Argentero, P., Munizza, C., &
Horne, R. (2009). The utility of the necessity-concerns framework in
explaining treatment non-adherence in four chronic illness groups in
Italy. Chronic Illness, 5, 129–133.
Unesco. (1997). ISCED 1997: International standard classification of
education. Retrieved from http://www.uis.unesco.org/publications/
van den Akker, M., Buntinx, F., Metsemakers, J. F. M., Roos, S., &
Knotterus, J. A. (1998). Multimorbidity in general practice: Prevalence,
incidence, and determinants of co-occurring chronic and recurrent dis-
eases. Journal of Clinical Epidemiology, 51, 367–375.
Ware, J. E., & Sherbourne, C. D. (1992). The MOS 36-item short-form
health survey (SF-36): I. Conceptual framework and item selection.
Medical Care, 30, 473–483.
Weinstein, N. D., & Rothman, A. J. (2005). Commentary: Revitalizing
research on health behavior theories. Health Education Research, 20,
World Health Organization. (2003). Adherence to long-term therapies:
Evidence for action. Geneva, Switzerland: World Health Organization.
Wroe, A. (2002). Intentional and unintentional nonadherence: A study of
decision making. Journal of Behavioral Medicine, 25, 355–372.
Wurm, S., Tomasik, M. J., & Tesch-Römer, C. (2010). On the importance
of a positive view on aging for physical exercise among middle-aged and
older adults: Cross-sectional and longitudinal findings. Psychology &
Health, 25, 25–42.
E-Mail Notification of Your Latest Issue Online!
Would you like to know when the next issue of your favorite APA journal will be available
online? This service is now available to you. Sign up at http://notify.apa.org/ and you will be
notified by e-mail when issues of interest to you become available!
CHANGES IN MEDICATION BELIEFS