1 BMQ-Original
1 BMQ-Original
1 BMQ-Original
To cite this article: Robert Horne , John Weinman & Maittew Hankins (1999) The beliefs
about medicines questionnaire: The development and evaluation of a new method for
assessing the cognitive representation of medication, Psychology and Health, 14:1, 1-24, DOI:
10.1080/08870449908407311
This paper presents a novel method for assessing cognitive representations of medication: the Beliefs about
Medicines Questionnaire (BMQ). The BMQ comprises two sections: the BMQ-Specific which assesses repre-
sentations of medication prescribed for personal use and the BMQ-General which assesses beliefs about mcdi-
cines in general. The pool of test items was derived from themes identified in published studies and from
interviews with chronically ill patients. Principal Component Analysis (PCA) of the test items resulted in a log-
ically coherent. 18 item, Cfactor structure which was stable across various illness groups. The BMQSpecific
comprises two Sitem factors assessing beliefs about the necessity of prescribed medication (Specific-Necessity)
and concerns about prescribed medication bascd on beliefs about the danger of dependence and long-term toxi-
city and the disruptive effects of medication (Specific-Concern). The BMQGeneral comprises two Citem fac-
tors assessing beliefs that medicines are harmful, addictive, poisons which should not be taken continuously
(Geneml-Horn) and that medicines arc overused by docton (General-Overure). The two sections of the BMQ
can be used in combination or separately. The paper describes the development of the BMQ scales and presents
data supporting their reliability and their criterion-related and discriminant validity.
KEY WORDS: Medicines. attitudes. personal models, illness pcmptions, drug therapy. treatment adherence.
INTRODUCTION
The prescription of a medicine is the most common treatment intervention and accounts
for the largest single commodity source of health expenditure in most developed
economies. However, it is estimated that approximately 30-5096 of prescribed medica-
tion is not taken as directed (Meichenbaum and Turk, 1987) and non-adherence to med-
ication is seen as a significant challenge to research and practice within the health care
domain (Home, 1993; Howitz and Horwitz. 1993). Various social cognition models
(SCMs) such as the Health Belief Model (HBM: Rosenstock, 1974). the Theory of
Reasoned Action (TRA: Ajzen and Fishbein, 1980) and its revision the Theory of
* Corresponding author.
1
2 R. HORNE ETAL
Planned Behaviour (TPB: Azjen, 1985), have been used to explain variation in medica-
tion adherence. This research shows that medication non-adherence may be the result of a
rational decision by the patient and identifies some of the cognitions which are salient to
these decisions. Although the specific type of beliefs which are associated with adherence
varies across studies, certain cognitive variables included in SCMs appear to be prerequi-
sites of adherence in some situations (Home and Weinman, 1998). For example, beliefs
that failure to take the treatment could result in adverse consequences and that one is per-
sonally susceptible to these effects tends to be associated with higher adherence rates
(e.g. Cummings et al., 1981; Kelly et al., 1987). Additionally, adherence decisions may
be influenced by a cost-benefit analysis in which the benefits of treatment are weighted
against the perceived barriers (e.g. Brownlee-Duffeck et al., 1987; Cummings e f al.,
1981). Other studies, based on the TRA/TPB have shown that the perceived views of sig-
nificant others such as family, friends and doctors (normative beliefs) may also influence
adherence (Cochran and Gitlin, 1988; Ried and Christensen, 1988; Ried et al., 1985).
Leventhal’s self-regulatory model of illness (SRM) (Leventhal et al., 1980; Leventhal
and Cameron, 1987) has also been applied to the study of medication adherence. In the
SRM the decision about whether or not to take medication is conceptualised as one of a
number of possible procedures for coping with an illness threat (Leventhal et al., 1997).
Adherence will be more likely if the patient perceives that the advice to take medication
makes ‘common-sense’, in the light of their experiences (e.g. past illness and/or current
symptoms) and their personal beliefs about the illness (Leventhal et al., 1992). In addi-
tion to providing an explanatory framework for how beliefs and behaviour are related,
self-regulatory theory postulates the types of beliefs which underpin illness cognitions
suggesting that the selection of a coping procedure, e.g. to seek (or not to seek) medical
advice or to take (or not to take) medication, is guided by beliefs about the nature, dura-
tion, causes, consequences and potential for cundcontrol of the illness.
It has been suggested that representations of treament may also play a role in self-reg-
ulation and that the explanatory power of SCMs in relation to medication adherence may
be enhanced by assessing patients’ beliefs about medication. Decisions about taking med-
ication are likely to be informed by beliefs about medicines as well as beliefs about the
illness which the medication is intended to treat or prevent (Home, 1997). This principle
is recognised in a recent report from the Royal Pharmaceutical Society of Great Britain
which has identified the role of medication beliefs in treatment adherence as a priority for
future research (Marinker, 1997; Royal Pharmaceutical Society of Great Britain, 1997).
Several qualitative studies have shown that people have beliefs about medicines in
general (e.g. Britten, 1994; Fallsberg, 1991 and Lorish et al., 1990). as well as beliefs
about medication prescribed for specific illnesses such as epilepsy (Conrad, 1985) and
hypertension (Morgan and Watkins, 1988). Moreover, certain representations of medi-
cines appear to be common across several illness and cultural groups. However, a system-
atic comparison of findings is hampered by the fact that the few studies which have
quantitatively assessed medication beliefs have used different questionnaires (Woller
et al., 1993; Echabe et al., 1992) or have investigated medication beliefs in the broader
context of views about the practice of medicine (Marteau, 1990). Furthermore, some
studies have assessed peoples’ ideas about medicines in general (General beliefs)
whereas others have focused on specific medication prescribed for a particular illness
(Specific beliefs).
A review of the existing literature on lay beliefs about medicines raises three key ques-
tions (Home, 1997). The first relates to the nature of medication beliefs and whether the
BELIEFS ABOUT MEDICINES QUESTIONNAIRE 3
range of specific and general medication beliefs can be summarised into ‘common themes’
which are relevant across illness and cultural groups. A second question relates to the dis-
tribution of these beliefs (who holds them and how strongly are they held?). Finally, there
is the question of how representationsof medicine relate to each other (e.g. general vs. s pe
cific) and to illness beliefs. as well as to adherence behaviours. We believe that there is
need for a psychometrically sound method for operationalisingand scoring commonly held
beliefs about medication in order to systematicallyaddress the above questions. This paper
describes the development of a questionnaire-based method for assessing beliefs about
Specific and General medication, the Beliefs about Medicines Questionnaire (Section 1)
and presents a preliminary evaluation of its psychometric properties (Section 2).
PARTICIPANTS
Brighton. UK, during a 3-month period between March and May 1994. Of 105 asthmatic
patients on the clinic schedule 17 refused to take part in the study, 9 did not attend the
clinic and one who agreed to take part subsequently withdrew without completing the
questionnaire. Seventy-eight patients entered the study and completed the clinic question-
naire giving an overall response rate 78/105 =74.3%. The mean duration of asthma was
1.6 years (SD = 1.3).
The Diabetic sample ( n = 99) comprised consecutive attenders at a diabetic out-patient
review clinic in a London general hospital. In a six week period, during April and May
1994, 124 study-eligible patients were approached and 20 refused to take part. Five of the
104 questionnaires returned were rejected ( > 10% of responses to questionnaire state-
ments were missing or illegible) giving a final completion rate of 79.8%. Sixty four
(64.7%) of the patients were insulin-treated while the remaining patients received oral
hypoglycaemic medication.
The Renal sample (n=47) was recruited from the renal unit at a London Teaching
Hospital. Patients were randomly selected from the dialysis list and evaluated for entry
into the study until a target sample of approximately half of the 103 patients on the hospi-
tal haemodialysis list were recruited. Of 59 randomly selected study-eligible patients,
47 agreed to take part and completed the questionnaire giving a response rate of 79.7%.
The mean duration of dialysis treatment was 4.5 years (SD=4.9).
The Psychiatric sample (n = 89) was recruited as part of an audit evaluating a medi-
cines-information service at a hospital psychiatric out-patient clinic in Brighton, UK. Of
118 patients who were eligible for inclusion in the present study, 27 failed to attend the
‘research clinic’ and a further two patients were omitted because they did not legibly
complete over 90% of the questionnaire items. The final study sample therefore com-
prised 89 patients giving a response rate of 78.4%. The mean duration of psychiatric ill-
ness was 10.2 years (SD= 8.4).
The Cardiac and General Medical inpatient samples ( n = 120; n =91 respectively) were
recruited from general medical wards of two London teaching hospitals and five district
general hospitals in London and Brighton, over an 8 week period between January and
March, 1995. Of 254 study-eligible patients, 37 refused to take part and 217 entered in the
study. Six of the questionnaires were rejected ( > 10% of responses to questionnaire state-
ments were missing or illegible). The remaining 211 questionnaires were retained for
analysis. The fmal completion rate was therefore 2 11/254= 83.1 %. On the basis of primary
diagnosis the sample comprised chronic cardiac disease (56.8%). chronic respiratory dis-
eases (16.2%), gastro-intestinal disorders (10.9%). diabetes (9.9%), cancer (3.8%) and
epilepsy (2.4%). Patients with chronic cardiac disease were considered as a single illness
group and the remaining patients ( n = 91) were grouped together as the ‘General medical
inpatients’.
METHOD
Components Analysis (PCA). Specific and General medication beliefs were analysed
separately. The factor structures obtained were then tested in three ways. Confirmatory
factor analysis (Tabachnick and Fidell, 1993) was used to verify the factor structure. The
stability of the factor structure across chronic illness groups was tested by investigating
whether the factor structure obtained by exploratory PCA in one illness group was repli-
cated in other illness groups. Finally, to confirm the validity of separating Specific and
General medication beliefs, items loading on the Specific and General factors identified
by PCA were combined and subjected to a further FCA. A high degree of separation
between general and specific items would indicate that patients made clear distinctions
between specific and general medication and justify the division of the BMQ into
Specific and General components.
Item Pool
A pool of 34 statements representing commonly held beliefs about specific (n= 16) and
general medication (n = 18) was obtained by selecting beliefs identified in the literature
which appeared to be common to patients with a range of chronic illnesses and from
interviews we conducted with 35 patients receiving regular medication for chronic illness
(20 haemodialysis patients and 15 patients with myocardial infarction). In these inter-
views patients were asked open questions eliciting their views about medicines pre-
scribed for them and their thoughts about medicines in general in an attempt to identify
common beliefs which had not emerged in previous studies. The final pool of 34 items,
together with their origin. is shown in Table 2. Twelve items were positive statements
about medicines (e.g. ‘Without medicines doctors would be less able to cure people’) and
the remaining 22 items focused on negative (e.g. ‘Most medicines are addictive’) or neu-
tral aspects (e.g. ‘Medicines only work ifthey are taken regularly’). This balance of items
reflects that observed in the literature (e.g. Britten. 1994; Donovan and Blake, 1992;
Fallsberg. 1991 and Lorish et al., 1990, Morgan and Watkins, 1988; Conrad, 1985) and
in interviews with patients. Responses to each statement were scored on a 5-point
Liken scale (where 1 = strongly disagree, 2 = disagree, 3 =uncertain, 4 =strongly agree
and 5 =strongly agree) and subjected to PCA as described below. Although the psychi-
amc and diabetic samples received identical General items as the other illness groups, the
Specific item pool differed by one item in the case of the psychiatric sample and two
items for the diabetic sample. This was done in order to reflect issues which were per-
ceived to be pertinent to these groups. For the psychiatric sample the item ‘Without my
medicines I would be very ill’ was replaced by ‘Only my medicines can control my mental
health problems’. For the diabetic sample the items ‘My life would be impossible without
my medicines’ and ‘My medicines protect me from becoming worse’ were replaced by the
items ‘My medication controls my diabetes’ and ‘My medication prevents my blood sugar
from becoming too high’. For this reason, the derivation of the BMQ-Specific scales was
based on data from the asthmatic, cardiac, renal and general medical in-patient groups
which had received identical Specific items.
PROCEDURE
Each participant was invited to take part in a study of patients’ views about their illness
and treatment, The investigators stressed that the study was being conducted by the
6 R. HORNE ETAL.
Soume of statements
University and was completely independent of the hospital and that responses were con-
fidential and anonymous and would not be seen by any of the staff involved in their care.
It was hoped that this would encourage participants to respond in a way which repre-
sented their own views rather than those which they considered to be socially desirable
BELIEFS ABOUT MEDICINES QUESTlONNAIRE 7
(Abraham and Hampson, 19%) and so avoid any response bias which might have resulted
if patients had associated the researcher with the clinical team. Participants were pre-
sented with the 34 item pool as described above at the same time as a battery of question-
naires assessing other relevant constructs as described in Section 2 below. These
measures (e.g. reported adherence and beliefs about illness) were included to assess the
criterion-related validity of the BMQ and were chosen on the basis of hypothesised rela-
tions with medication beliefs. The instructions to participants, are shown in the
Appendix. Clinic patients were asked to complete the questionnaire while waiting to see
the doctor. Patients recruited from hospital wards were asked to complete the study ques-
tionnaire by the researcher who then arranged to collect it at a convenient time.
RESULTS
Exploratory PCA
Specific beliefs. The mean and SD for each of the 16 items eliciting beliefs about pre-
scribed medication administered to the Cardiac sample are shown in Table 3.
Four items with KMO values cO.7 were omitted. Factor scree plot analysis suggested a
2-factor solution explaining 5 1% of the variance. Having arrived at a core structure of two
5-item factors the data set was cleaned by removal of multivariate outliers (Mahalanobis
distance > 3 standard deviations from the multi-dimensional mean) and removal of cases
with greater than five missing items. This resulted in omission of 6 cases. Re-factoring on
the 114 remaining cases produced a similar two factor structure explaining 53% of the
variance.
Factor labels. The final 2-factor structure is shown in Table 4. The first factor comprised
items relating to the positive effect of medication on health and were representative of the
perceived necessity of medication for maintaining health. This factor was labelled
Specific-Necessity. The second factor comprised items relating to concerns about the
adverse consequences of medication based on beliefs about the potential for dependence
or harmful long-term effects and that medication taking is disruptive. This factor was
labelled Specific-Concerns.
General beliefs. The mean and standard deviation for scores on each of the 18 items elic-
iting beliefs about medicines in general are shown in Table 3. Elimination of six items
with a low KMO statistic (<0.7) and setting a two factor solution as suggested by scree
plot analysis, followed by elimination of a further 4 items with low or diffuse loading
resulted in two 4-item factors shown in Table 5 .
BELIEFS ABOUT MEDICINES QUESTIONNAIRE 9
Table 3 Mean and slandard deviation SD of responses to specific and general statements
Mean SD
Staremets about prrscribcd medication (Spccijic)
It is difficult for me to take my medicines in exactly the way my doctor told me 2.09 0.75
My medicines disrupt my life 2.31 0.92
Having to take medicines worries me 2.70 1.07
I sometimes worry about becoming too dependent on my medicines 2.82 1.10
My medicines are a mystery to me 3.00 0.98
I sometimes worry about the long-term effects of my medicines 3.11 1.15
My medicines are powerful 3.33 0.77
I would like to change my prcsent treatment 3.44 1.01
My life would be impossible without my medicines 3.51 0.95
My health in the future will depend on my medicines 3.62 0.93
I can cope without my medicines 3.62 O.%
Without my medicines I would be very ill 3.66 0.88
1. am in control of my medication 3.73 0.85
My medicines protect me from becoming worse 3.91 0.71
My medicines are effective 3.94 0.56
My health, at prcsent. depends on my medicines 4.03 0.73
Without medicines doctors would be less able to cure people 3.13 1.54
Newer medicines are more effective than older ones 3.37 0.84
Most medicines are addictive 2.73 0.89
People who take medicines should stop their treatment for 2.54 0.9 I
a while every now and again
Medicines only work if they are taken regulary 3.75 0.80
Medicines do more harm than good 2.24 0.85
Medicines are not natural remedies 3.13 0.92
All medicines arc poisons 2.24 0.97
It is better to d o without medicines 2.61 1.08
Natural remedies are safer than medicines 2.88 0.91
Stronger medicines are mom dangerous than weaker medicines 3.24 0.90
Medicines are a necessary evil 3.06 1.10
Doctors place too much trust in medicines 2.90 0.93
If doctors had more time with patients they would prescribe fewer medicines 3.17 0.98
Then: is a big difference between a medicine and a drug 3.24 0.88
The medicine you get is more important than the doctor you see 2.87 1.14
Doctors use too many medicines 2.84 0.9 1
Most medicines are safe 2.72 0.92
The first factor comprised items expressing beliefs about the way in which medicines
are used by doctors. The essence of this factor, labelled General-Overuse is the notion
that medicines are over-prescribed by doctors who place too much trust in them. The sec-
ond factor, labelled General-Harm concerns the potential of medication to harm and
comprises representations of medication as harmful, addictive, poisons and the belief that
people who take medicines should stop their treatment every now again.
Table 5 Factor structure obtained by PCA of BMQGeneral items (n -219 patients with
chronic illnesses-asthmatic977. diabetic =99,haernodialysisrecipients -42)'
~ ~~~
If doctors had more time with patients, they would 0.80 0.11
prescribe fewer medicines
Docton use too many medicines 0.79 0.15
Docton place too much trust in medicines 0.72 0.24
Natural remedies are safer than medicines 0.70 0.33
Medicines do more harm than good 0.33 0.72
People who take medicines should stop their 0.18 0.70
treatment for a while every now and again
Most medicines an addictive 0.02 0.70
All medicines arc poisons 0.28 0.69
Eigenvalue 2.8 1.5
Percentage variance explained 35.3 19.0
*Five cues WQC removed during thc cleaning proctdurr.
BMQ-General
Overuse 0.90 NA NA 0.70 0.88 NA
Harm 0.93 NA NA 0.73 0.83 NA
BMQSpecific
Necessity 0.98 0.92 0.88 0.95 0.83 0.90
Concerns 0.98 0.88 0.88 0.90 O.% 0.95
Replication of factor strucfure. The 2-factor structure for Specific beliefs was replicated
by PCA of the responses to the 10 items obtained from asthmatic. renal and general med-
ical inpatient samples. Although there were minor differences in factor loadings, the fac-
tor structure obtained for each of the samples contained identical items. The 2-factor
structure obtained for General beliefs by exploratory PCA of combined data from the
asthmatic, diabetic and renal samples was replicated in the cardiac, and psychiatric sam-
ples, indicating acceptable stability of the factor structures across illness groups. PCA of
the data from the General Medical in-patients, produced a similar factor structure. with
the exception of one item: “Natural remedies are safer than medicines” which had
migrated from factor 1 to factor 2.
PCA of combined Specific and Generalfactor item. PCA of pooled data from all 6 illness
samples showed a clear separation of Specific and General items. A 4-factor structure was
obtained (see Table 8) which closely resembled the original Specific and General factor
structures except that one item from the Specific-Concernsfactor ‘My medicines are u mys-
tery to me’, loaded a little higher on the General Harm (0.55) than on Specific-Concerns
(0.39). Removal of the General Medical Inpatient sample from the data set followed by a
further PCA on pooled data from the discreet diagnostic groups (asthmatic, diabetic, renal,
cardiac and psychiatric) replicated the original Specific and General factor structures.
PARTICIPANTS
Table 8 Structure matrix obtained by PCA on combined items from the Specific and General medication
belief factors on pooled data from the six illness groups comprising the main sample (total n = 524)
( > 90% items answered legibly). The response rate for the Allopathic Care sample was
therefore 104/126 = 83%. The Complementary Care sample were recruited from the
clinics of a single herbalist and single homeopath, in Brighton, during the same time
period as the Allopathic Care sample. Both practitioners felt that it would be inappro-
priate to base a researcher in the clinic and so patients were invited to take part in the
study by the herbalist/homeopath. Those who agreed were asked to fill out the ques-
tionnaire and return it to the author at the University of Brighton in the stamped
addressed envelope provided. Fifty-four questionnaires were given out and 36 com-
pleted questionnaires were returned. The final response rate for the Complementary
Care sample was therefore 36/53 = 67.9%.
Matched samples. Seventy two participants were matched for age and sex and educa-
tional experience. Patients from the Allopathic Care sample were selected to match the
age and gender profile of the Complementary Care group. Matching was canied out
because of the large disparity in group sizes and the possible confounding effect of age
and gender. The characteristics of the matched samples are shown in Table 9.
There were no significant differences between Allopathic and Complementary samples in
terms of age, and gender. The Complementary Sample had significantly greater educational
experience (Pearson Chi-square =6.34; DF =2; p ~ 0 . 0 5and
) had made significantly more
visits to homeopathic (r=3.35; n=72; p<O.OOl) and herbal (t=4.84; n=72; p<O.OOI)
BELIEFS ABOUT MEDICINES QUESTIONNAIRE 13
n 36 36
Age [mean (SD)] 42.3 (11.1) 47.3 (18.6)
Gender: number (5%) male 9 (25) 8 (22)
Educational Expcrience
Secondary (5%) 66.6 44.4
Tertiary (5%) 16.7 16.7
Advanced (%) 16.7 38.9
Mean (SD)number of visits over
previous 6 months to:
General practitioner 2 (1.8) 1.7 (1.9)
Homeopath 0.03 (0.17) 0.78 (1.33)
Herbalist 0 1.5 (1.9)
Mean (SD)N u m b of hospital admissions 0.36 (1.1) 0.19 (0.58)
over previous year
practitioners in the 6 months prior to the study than had the Allopathic Care sample. There
were no significant Werences between the samples in the number of reported visits to
NHS General Practitioners or hospital admissions. The latter finding was interpreted as an
indicator that the samples were comparable in terms of illness severity.
MEASURES
The Illness Perception Questionnaire (IPQ) (Weinman et al., 19%). The IPQ comprises
five scales measuring the five components of illness representation specified in
Leventhal’s self-regulatory model of illness (Leventhal et al., 1980). The five scales
assess identify (the symptoms the patient associates with the illness), cause (personal
ideas about aetiology), time line (the perceived duration of the illness), consequences
(expected effects and outcome), and curelcontrol (beliefs about potential for cure and
control of the illness). The psychometric properties of the IPQ have been evaluated in
7 patient groups including asthmatic, diabetic and hospital haemodialysis recipients and
the internal consistency, test-retest reliability and the concurrent, discriminant and pre-
dictive validity of the IFQ scales are within acceptable limits (Weinman et al., 1996).
Reported Adherence to Medication (RAM) scale. Published adherence self-report scales
were thought to be unsuitable because they are not specific to medication (DiMatteo
et al., 1993; Kravitz et al., 1993) or because they do not elicit self-report of the fre-
quency of adjusting or altering dosages (Morisky, 1986). A reported adhecence to med-
ication scale (RAM)was therefore devised for the present study. Non-adherence was
indicated by the tendency to forget to take medication and to deliberately adjust or alter
the dose from that recommended by the physician. The RAM scale comprises four
adherence statements. Two items (‘I sometimes forget to take my medicines’ and
‘I sometimes alter the dose of my medication to suit my own needr’) are scored on a
5-point Likert scale with reverse scoring (where 1 =strongly agree; 2 = agree; 3 = uncer-
tain;4 =disagree and 5 =strongly disagree). A further two items (‘Somepeople forget to
take their medicines. How often does this happen to you?’ and ‘Somepeople I have talked
to say that they miss out a dose of their medication or adjust it to suit their own needr.
How ojien do you do this?’) are phrased as direct questions asking the patient to report
14 R. HORNE EFAL.
’hblc 10 Items assessing medication-related cognitions used for psychometric evaluation of the BMQ scales
tivity to the potential adverse effects of medication (e.g. ‘Even small amounts ofmedi-
cines can upset my body’). The scale is currently under development at Rutgers
University New Jersey, USA (Diefenbach et al., 1997) and details of scale items are
available from the authors. Responses are scored on a 5-item Liken scale and the indi-
vidual item scores are summed to give a total Sensitive Soma score ranging from 5 to 25
where high scores=high perceived sensitivity to the potential adverse effects of medica-
tion: This Sensitive Soma scale was administered to the cardiac ( n = 120) and general
medical in-patient (n = 91) samples. The internal reliability of the scale, as measured by
Cmnbach’s alpha, was acceptable in both groups (general-medical=0.80;cardiac =0.78).
0 Single measures assessing medication-related cognitions. The psychometric evaluation
of the BMQ utilised three of the single item statements from the original 34-item pool
described above. The items had not loaded on the BMQ factors and so did not represent
a Specific-Necessity, Specific-Concern, General-Harm or General-Overuse cognition.
However, they seemed, at face value, to represent interesting medication related cogni-
tions and so were used for psychometric evaluation of the BMQ scales.
In addition to these items a further two single item statements were also included as
shown in Table 10. Responses to all five single items were: scored on a 5-point Likert
scale where 1 = strongly disagree and 5 =strongly agree.
Criterion-relatedvalidity
The assessment of the criterion-related validity of each of the BMQ scales was based on
the following predictions:
1. Specific-Necessiry. Patients with stronger beliefs in the necessity of their medication
would be less likely to believe that they can cope without it. Thus scores on the
BELIEFS ABOUT MEDICINES QUESTIONNAIRE 1s
Specific-Necessity scale would be negatively correlated with scores on the item: ‘I can
cope without my medicines’. Beliefs in the necessity of prescribed medication would
also be related to perceptions of illness. In particular, patients who believed that their
illness would last a long time and who experienced more symptoms would have
stronger beliefs in the necessity of the medication prescribed to treat it. Thus
Specific-Necessity scores would be positively correlated with scores on the Identity
and Timeline components of the IPQ which respectively assess perceptions of symp-
tom severity and likely duration of the illness.
2. Specific-Concern. Patients with stronger concerns about their prescribed medication
would be more distrustful of it, would tend to want more information about it and
would be more likely to want to change their current treatment. Thus it was hypothe-
sised that the SpeciJic-Concern scale scores would be positively correlated with scores
on the ‘Lackof trust in prescribed medication’ and ‘Desire to change present treatment’
items and would be negatively correlated with scores on the ‘Satisfaction with amount
of medicines information received’ item. Additionally, those who perceived themselves
to be susceptible to the potential adverse effects of medication would have stronger
concerns about their prescribed medication. Thus scores on the Specific-Concern scale
would be positively correlated with scores on the Sensitive Soma scale.
3. General-Ham. Patients who believed that medicines in general are intrinsically harm-
ful would be more likely to believe that it is better to avoid taking them. Thus scores
on the General-Ham scale would be positively correlated with scores on the ‘It is bet-
ter to do without medicines’ and ‘I can cope without my medicines’ items. Moreover,
participants who believed that medicines in general are intrinsically harmful would be
more likely to consider themselves to be susceptible to potential adverse effects of
medication. Thus scores on the General-Harm scale would be positively correlated
with scores on the Sensitive Soma scale which assess perceptions of personal sensitiv-
ity to the adverse effects of medication.
4. General-Overuse. Scores on the General-Overuse scale would be positively correlated
with scores on the ‘Ican cope without my medicines’ and the ‘It is better to do without
medicines’ items.
5 . Relations between BMQ scales and reported adherence to medication (RAM). It was
hypothesised that stronger beliefs in the necessity of prescribed medication would be
associated with higher reported adherence. Thus, Specific-Necessify scores would be
positively correlated with the RAM scale scores. Conversely, patients with stronger
concerns about prescribed medication and those who believed that medicines in general
were harmful substances which are overused by doctors would report lower medica-
tion adherence rates. Thus correlations between the Specific-Concerns, General-Harm
and General-Overuse and the RAM scale would be negative.
Discriminant Validity
The discriminant validity of the BMQ-Specific scales was tested on the basis of their
ability to distinguish between different illnesses and hence treatment modalities. The dis-
criminant validity of the BMQ-General scales was tested on the basis of their ability to
distinguish between patients presenting a personal prescription at a community pharmacy
and those seeking complementary therapies. The specific hypotheses were as follows:
1. SpeciJic-Necessity.Beliefs about the necessity of prescribed medication would be influ-
enced by the type of treatment typically prescribed for the illness. The characteristic
16 R. HORNE ETA5
PROCEDURE
The psychometric evaluation was conducted on the basis of interactions between the
BMQ factors and the above measures which had been administered to the main sample at
the same time as the pool of mediation belief items from which the BMQ was derived.
The AllopathidComplementary Care samples were recruited after the BMQ had been
derived fmm the main sample (as detailed in Section I). Only the 8-item BMQ-General
(comprising the General-Overuse and General-Harm scales) was administered to the
Allopathic/Complementary Care samples. The Sensitive Soma Scale was not available
when the asthmatic, diabetic, renal and psychiatric samples were recruited. The scale was
however available when the cardiac and general medical samples were recruited a few
months later. Thus different samples were used to evaluate different psychometric proper-
ties. The internal reliability of each scale was evaluated for all 6 illness groups compris-
ing the main sample. Test-retest reliability was evaluated using the asthmatic sample.
Repeat questionnaires were sent to the patients, together with a stamped addressed enve-
lope, two weeks after they had been seen in clinic. Criterion-related validity of the
BMQ-Specific scales was evaluated using the asthmatic sample, except for interaction
between the Specific-Concernsand Sensitive Soma scales which were evaluated using the
BELIEFS ABOUT MEDICINES QUESTIONNAIRE 17
general medical inpatient samples. Relations between BMQ scales and RAM were evalu-
ated on pooled data from the Cardiac and General Medical samples. The discriminant valid-
ity of the BMQ-Specific scales was evaluated in the main sample. The discriminant validity
of the BMQ-General scales was evaluated in the Allopathic/Complementary Care sample.
Stutisrical Techniques
The internal consistency of each BMQ scale was evaluated using Cronbach's alpha.
Spearman correlations (p) were used to evaluate test-retest reliabilities between initial
and repeated test scores for each scale and also the relations between scales used to test
the criterion-related validity of the BMQ. The a priori hypotheses relating to the discrim-
inant validity of the BMQ-Specific scales were investigated using one-way ANOVA and
linear contrasts. Further differences between illness samples were identified using (post
hoc) Tukey's HSD test. Multivariate analysis of variance (MANOVA) was not used for
analysis of differences in measures due to the moderate level of intercorrelation between
Specific-Concernsand General-Ham (p = 0.31; n =524; p c 0.01) and General-Overuse
(p=O.24; n=524; pc0.01).Differences in mean BMQ-General scores between Allo-
pathic and Complementary care seekers was assessed using an independent samples
t-test. A one-tailed test was used as the direction of association had been specified within
the relevant hypothesis.
RESULTS
a b l e 11 Internal consistency (Cronbach alpha) for the BMQ scales and test-retest correlations
returned the repeat questionnaires, giving a 40% response rate. The correlation coeffi-
cients shown in Table 11 indicate that the test-retest reliability of the scales is within
accepted limits. Correlations between BMQ scales are shown in Table 12.
Criterion-related Validity
Specific-Necessity. Evidence for the criterion-related validity of the Specific-Necessity
scale was provided by the negative correlation between scale scores and responses to
the statement: “ I can cope without my medicines” (p = -0.44; n = 78; p <0.001) as
expected. As predicted there were also positive correlations with scores on the IPQ
Timeline (p =0.49; n = 77; p <0.001) and Identity (p =0.24; n = 76; p <0.05) scales
which measure perceived duration and subjective symptomatology of the illness.
Specific-Concerns. Scores for the asthmatic group were positively correlated with the
statement: ‘Icannot always trust my medicines’ (p =0.33; n =78; p < 0.005), and ‘I
would like to change my present treatment’ (p =0.37; n =78; p (0.001). The hypothe-
sis that Specific-Concerns would be associated with a desire for more information
about medicines was confirmed by the significant negative correlation with responses
to the statement: ‘I have been given enough information about my medicines’
(p = -0.45; n =78; p <0.001). As hypothesised, a significant positive correlation was
obtained between Specvc-Concerns and beliefs about personal sensitivity to the
adverse effects of medication as assessed by the Sensitive-Soma scale administered to
the General Medical and Cardiac samples (p = 0.5. n = 2 1 1, p <0.00 1).
General-Ham and General-Overuse. Correlation between General-Harm scores and
responses to the single item statement “It is better to do without medicines” was as
expected (p = 0.23; n = 78; p ~ 0 . 0 5 )Responses
. to the statement “ I can cope without
my medicines” correlated significantly, in the predicted direction, with both the
General-Ham (p = 0.24; n = 77; p <0.05) and General-Overuse scales (p = 0.34;
n = 78; p <0.005). Correlations between the General-Ham and Sensitive-Soma scales
(p = 0.25, n = 9 1, p <0.05), although small in magnitude, were in the predicted direc-
tion and statistically significant.
Adherence to treafment. Correlations between BMQ scales and reported adherence
assessed by the RAM scale to medication were as expected. Specific Necessity beliefs
correlated with higher reported adherence (p =0.19; n =210, ~ ~ 0 . 0 1Correlations
).
between the RAM scale and the Specific-Concerns (p= -0.28; n=210; pc0.001).
General-Overuse(p= -0.19; n=210;p<0.01) and General-Ham (p= -0.06;n=210;
p > 0.05) scales were! all in the predicted direction, although those between the RAM
and General-Harm scales failed to reach statistical significance.
Discriminant Validity
1. BMQ-Specific scales. Table 13 shows the results of a series of one-way analyses of
variance (ANOVA), with (a priori) linear contrasts and (post-hoc)Tukey’s HSD tests
SpecificConcems -0.01
General-Harm -0.05 0.3 1*
General-Overuse -0.17 0.24* 0.40.
BELIEFS ABOUT MEDICINES QUESTIONNAIRE 19
Table 13 Scale means and standard deviations for BMQ scales for the six illness groups comprising the
main sample
Specific-Necessity
Mean 19.67, 21.26, 19.4Sb, 18.7&, 17.72, 19.65, 11.73 ~0.01
SD 3.23 2.98 2.78 3.02 3.75 3.92
Specific-Concerns
MKUl 15.76, 12.91, 13.77, 13.95, 15.60,, 14.26, 7.49 ~0.01
SD 4.09 3.38 4.28 3.73 3.36 3.92
General-Harm
Meall 10.24, 9.29, 9.91, 9.98, 9.92, 9.86, 1.29 0.26
SD 2.30 2.43 3.76 2.32 2.81 2.80
General-Ovemsc
M a 11.64,, 11.43, 12.66,,, 12.80, 2.25,b 12.42,,, 3.48 0.01
SD 2.59 2.77 3.19 2.90 2.84 2.76
Note: Means sharing a common suburipc arc no( significantly diffmnc by (0 prion) linear conuas8s (x (posf hoc) Tukcy’s
HSD lest ( p 20.05.).
Table 14 Group differences in BMQ-General scores for matched samples of orthodox and
complementary patients
~ ~- -
Measure Alloparhic Complenvnrary r P
( n = 36) (n = 36) (4-
70) (I-railed)
in which mean scores on the BMQ scales were compared across illness samples. It can
be seen that the BMQ scales were able to distinguish between patients on the basis of
illness (and treatment) groupings. The predictions for discriminant validity of the
Specific-Necessity scale were confirmed by the finding that diabetic group had signifi-
cantly higher Specific-Necessity scores than all other groups and the asthmatic patients
had significantly higher scores than the psychiatric outpatients who attained the lowest
mean as predicted. As was expected, the asthmatic and psychiatric samples had signif-
icantly higher Specific-Concernsthan the other illness groups, supporting the discrim-
inant validity of this scale.
2. BMQ-General scales. As was predicted, patients attending a Complementary clinic
(homeopath/ herbalist) had significantly higher scores on both the General-heme
and General-Harm scales than those presenting a personal prescription for dispensing
at a community pharmacy, as shown in Table 14.
20 R. HORNE ETAL.
DISCUSSION
Acknowledgements
This research was supported by a research award from the Pharmacy Enterprise Scheme,
Department of Health, UK.We would also like to acknowledge a number of colleagues
who made important contributions to the development of this questionnaire. Early discus-
sions with Professor Marie Johnston helped to focus our ideas. Thanks are due to
Professor Howard Leventhal and colleagues for permission to use the Sensitive Soma
scale and to Associate Professor Keith Petrie for helpful discussions. Railton Scott,
Angela Lashau, Barry Jubraj and Carol Kirkman helped with data collection from the
asthmatic, diabetic, renal and psychiatric illness groups. Thanks are also due to Alice
Ward, Linda Dodds and the pre-registration pharmacists of South East Thames who col-
lected data from the cardiac and general medical groups. We are also grateful to Professor
Richard Vincent and Drs Michael Rosenberg, John Hartley, Charles Turton (Btighton)
and to Dr David Taube and Sian Sumner (London) for their help and to those clinicians
who provided access to their patients and to the patients who took part. We would also
like to thank the two anonymous reviewers for their helpful comments.
References
Abraham, C.. Shecran, P.. Spears, R. and Abrams, D. (1992) Health beliefs and the promotion of KIV-
preventive intentions among teenagers: a Scottish perspective. Heulrh Psychology 11.363-370.
Abraham. C. and Hampson. S.E. (1996) A social cognition approach to health psychology: philosophical and
methodological issues. Psychology and Health 11.223-241.
22 R. HORNE E T L
Ajzen. 1. (1985) From intentions to actions: a theory of planned behaviour. In: Action-Conrml: Fmm Cognition
ro Behuviour. ed. Kuhl. J. and Beckmann, J. Heidelberg: Springer-Veriag.
Ajzen. I. and Fishbcin, M. (1980) Understanding anirudes and Predicting social behuviour, Englewood Cliffs.
NJ: Rentice-Hall.
Becker. M.H.. Radius, S.M., Rosenstock. LM., Drachman, R.H., Schubert, K.C. and Tetts. K.C. (1978)
Compliance with a medical regimen for asthma: a test of the health belief model. Public Healrh Repons 93,
268-277.
Britten. N. (1994) Patients’ ideas about medicines: a qualitative study in a general practice population. Brirish
J o u m l of General Practice 44,465-468.
Brownlee-Duffeck. M.. Peterson, L.. Simonds. J.F.. Goldstein, D.. Kilo. C. and Hoe@. S. (1987) The role of
health beliefs in the regimen adherence and metabolic control of adolescents and adults with diabetes melli-
tus. J o u m l of Consulring and Clinical Psychology 55. 139-144.
Cameron. L., Leventhal. E.A. and Leventhal, H. (1993) Symptom representations and affect as determinants of
cam seeking in a communitydwelling. adult sample population. Hedth Psychology 12. 171-179.
Cattell. R.B. (1995) The fallacy of five factors in the personality sphere. The Psychologist 8,207-208.
Conner. M. and Norman. P. (1996) Pndicring hcalrh bchuvwur: research and practice wirh social cognition
models, Buckingham: Open University Press.
Cochran, S.D. and Gitlin. M.J. (1988) Attitudinal correlates of Lithium compliance in bipolar affective disor-
ders. The J o u m l of Nervous and Mental Diseuse 17603): 457-464.
Conrad, P. (1985) The meaning of medications: another look at compliance. Social Science in Medicine 20(1):
29-31.
Cummings, K.M., Becker. M.H.. Kirscht, J.P. and Levin, N.W. (1981) Intervention strategies to improve com-
pliance with medical regimens by ambulatory haemodialysis patients. Jouml of Behuvioural Medicine 4.
11 1-127.
Diefenbach. M., Leventhal. H. and Leventhal. E.A. (1996). The Sensitive Soma Scale. Manuscript in prepara-
tion.
DiMatteo. M.R., Sherboume, C.D.. Hays, R.D.. Ordway, L.. Kravitz, R.L., McGlynn, E.A.. Kaplan, S. and
Rogers, W.H. (1993) Physicians’ characteristics influence patients’ adherence to medical treatment: results
from the Medical Outcomes Study. Healrh Psychology 12.93-102.
Donovan. J.L. and Blake, D.R. (1992) Patient noncompliance: Deviance or reasoned decision-making? Social
Science in Medicine 34. 507-5 13.
Echabe, A.E.. Guillen. C.S. and Ozamis J.A. (1992) Representations of health illness and medicines: coping
strategies and health promoting behaviour. British J o u m l of Clinical Psychology 31.339-349.
Fallsberg. M. (1991) Reflections on medicines and medicarion: 4 quolirative analysis among people on long-
term drug regimens, Linkbping. Sweden: Linkoping University.
Home. R. (1993) One to be taken as directed:reflections on non-adherence (noncompliance). l o u d of Social
and Adminisrrarive Phanocy 10, 150-156.
Home, R. (1997) Representations of medication and treatment: advances in theory and measurement. In:
Perceprions of Hedrh and Illness: Cumnr Research and Applications. ed. Peme. K.J. and Weinman, J.
London: Harwood Academic. pp. 155-187.
Home, R. and Weinman. J. (1998) Predicting treatment adherence: an overview of $coretical models. In:
Adherence to Treamenr in Medical Condirions, ed. Myers. L. and Midence. K. London: Harwood Academic
pp. 25-50.
Honvitz, R.I., Viscoli. C.M., Berkman. L., Murray, C.J.. Ransohoff, D.F. and Sindelar. J. (1990) Treatment
adherence and the risk of death after a myocardial infarction. h c e r 336,542-545.
Honuitz, R.I. and Honvitz, S.M. (1993) Adherence to treatment and health outcomes. Archives of Internal
Medicine 153. 1863-1868.
Janz. N.K.and Becker, M.H. (1984) The health belief model: a decade later. Healrh Mucarion Quonerly 11,
1-47.
Kelly. G.R., Mamon, J.A. and Scott, J.E. (1987) Utility of the health belief model in examining medication
compliance among psychiaeic outpatients. Social Science and Medicine 25, 1205-121 I .
Nine, P. (1994) An Easy Guide to Furor Anulysis, London: Rwtledge.
Kraviu. R.L.. Hays. R.D.. Shcrboume. C.D.. DMatteo, M.R.. Rogers, W.H., Ordway. L. and Greenfield, S.
(1993) Recall of recommendations and adherence to advice among patients with chronic medical conditions.
Archives of I n r c m l Medicine 153, 1869-1878.
Leventhal, H.. Meyer. D. and Nmnz, D. (1980) The common sense representation of illness danger. In
S. Rachman (Ed.) Conm’burionsro Medical Psychology. Oxfod. Pergamon Press.
Leventhal. H. and Cameron, L. (1987) Behavioural theories and the problem of compliance. Patient Mucarion
and Counselling 10, 117-138.
Leventhal. H.. b t u h g , D.V., Coons, H.L., Luchtehand, C.M. and Love, R.R. (1986) Adaptation to
chemotherapy treatments. In: Andersen. B.L. (Ed.) Women with Cuncrr: Psychological Perspectives.
pp. 172-203. New York: Springer Verlag.
BELIEFS ABOUT MEDICINES QUESTIONNAIRE 23
Leventhal. H. and Diefenback, M. (1991) The Active Side of Illness Cognition. In: Skelton, J.A. and
Croyle, R.T. (Eds.) Menrul Representorion in Heulrh and Illness, pp. 245-27 I. New York Springer Verlag.
Leventhal. H., Diefenbach. M. and Lcventhal. E.A. (1992) Illness cognition: using common sense to understand
treatment adherence and affect cognition interactions. Cogniriwc Therapy and Rcseurch 16(2). 143-163.
Leventhal, H..Benyamini, Y., Brownlee, S.. Diefenbach. D.. Leventhal, E.A., Pntrick-Miller. L. and Robitaille,
C. (1997) Illness representations: theoretical foundations. In: Perccprions of Heulrh and Illness: C u m n r
Reseurch nnd Applicurions. ed. PeUie, K J . and Weinman. J. London: Hanvood Academic, pp. 19-46.
Lorish. C.D.. bchards. B. and Brown. S. (1990) Perspective of the patient with rfreumatoid arthritis on issues
related to missed medication. Anhriris Cow and Research 3.78-84.
Marinker, M. (1997) From compliance to concordance: achieving shared goals in medicine taking. Brirish
Medic41 J o u ~ l 3 1 4 . 7 4 7 - 7 4 8 .
Marteau, T.M. (1990) Attitudes to doctors and medicine: the preliminary development of a new scale.
Psychology and Heulrh, 4,351-356.
Marteau. T.M. (1995) Health Beliefs and attributions. In: Broome, A. and Llewellyn. S . (Eds.) Heulrh
Psychology: Processes ond Applicutions. 2nd edn. pp. 3-20. London: Chapman and Hall.
Meyer, D. and Leventhal, H. (1984) Common sense models of illness: the example of hypertension. Heulrh
Psychology 4 2 ) . 115-135.
Meichenbaum. D. and Turk, D.C. (1987) Fociliruring rreurmenr adherence: 4 prucririoncr 's hundbook. New
York: Plenum Press.
Morgan, M. and Watkins, C.J. (1988) Managing hypertension: beliefs and responses to medication among cul-
tural groups. Sociology of Heulrh und Illness 10,561-578.
Morisky, E.E., Green, L.W.and Levine, D.M. (1986) C o n c m n t and predictive validity of a self-reported mea-
sure of medication adherence. Medical Cure 24.67-74.
Moss-Morris,R.. Petrie. K.J.and Weinman. J. (1996) Functioning in Chronic Fatigue Syndrome: do illness per-
ceptions play a regulatory role? Brirish Journul of Heulrh Psychology 1. 15-25.
Norusis, M.J. (1992) SPSS for Windows: Profession01 Srurisrics (Releare 5). Chicago: SPSS Inc.
Petrie. K.J..Weinman. J.. Sharpe. N. and Buckley. J. (1996) Rcdicting return to work and functioning following
myocardial infarction: the role of the patient's view of their illness. Brirish Medic41 J o u m l 3 1 2 . 1191-1 194.
Redeimcier, D.A.. Rozin, P. and Kahneman. D. (1993) Understanding patients' decisions: cognitive and emo-
tional persptctives. Journal of rhe Amricun Medicul Associclrion 270 (1). 72-76.
R i d . L.D. and Christensen, D.B. (1988) A psychosocial perspective in the explanation of patients' drug-taking
behaviour. Social Science in Medicine 27, 277-285.
R i d . L.D., Oleen, M.A.. M d n s o n . O.B. and Pluhar, R. (1985) Explaining intention to comply with antihyper-
tensive regimens: the utility of health beliefs and the theory of reasoned action. Joumul of Social und
Admisrruriwc Phurmory 3(2): 42-52.
Rosenstock, I. (1974) The health belief model and preventative health behaviour. Heulrh Uucurion
Monogruphs 2,354-386.
Royal Pharmaceutical Society of Great Britain (1997) From compliance ro conconirmcc: uchieving shared
g a l s in medicine ruking. London.
Skelton. J.A. and Croyle. R.T. (1991) Menrul Represenrution in Health and Illness, New York: Springer-Verlag.
Srrciner. D.L. (1994) Figuring out factors: the use and misuse of factor analysis. CoMdion Journal of
PSyChiotry 39(3): 135-140.
Tabachnik, B.G. and Fidell. L.S. (1989) Using multiwuri~resrurisrics. 2nd edn. New York: Harper Collins.
Weinman, J., Petrie. KJ., Moss-Moms, R. and Home, R. (1996) The Illness Perception Questionnaire: a new
method for assessing cognitive representations of illness. Psychology and Heulrh 11,43 1-445.
Woller. W., Knrse, J., Winter, P. and Mans, E J . (1993) Cortisone image and emotional support by key figures in
patients with bronchial asthma: an empirical study. Psychorherupy and Psychosomatics 59. 190-196.
RMQ-Specific
Your views about medicines prescribed for you*
We would like to ask you about your personal views about medicines prescribed for you.
These are statements other people have made about their medicines.
Please indicate the extent to which you agree or disagree with them by ticking the
appropriate box.
0 There are no right or wrong answers. We are interested in your personal views.
24 R. H O W ET AL.
Note:
To elicit beliefs about individual components of the treatment regimen the reference state-
ment should refer to the medicine by name e.g. Your views abour aspirin prescribed for you
Additionally items can refer to a named illness e.g. Your views about medicines pre-
scribed for your asthma
BMQ-General
Your views about medicines in general
We would like to ask you about your personal views about medicines in general.
These are statements other people have made about medicines in general.
Please indicate the extent to which you agree or disagree with them by ticking the
appropriate box.
There are no right or wrong answers. We are intersted in your personal views.