American Journal of Gastroenterology
C ?2007 by Am. Coll. of Gastroenterology
Published by Blackwell Publishing
Predictors of Medication Adherence in Inflammatory
Jason P. Ediger, Ph.D.,1,2John R. Walker, Ph.D.,1,2Lesley Graff, Ph.D.,1,2Lisa Lix, Ph.D.,1,2Ian Clara, Ph.D.,1,2
Patricia Rawsthorne, R.N.,1,3Linda Rogala, R.N.,1,3Norine Miller, R.N.,1,3Cory McPhail, B.A.,1,2
Kathleen Deering, B.A.,1,2and Charles N. Bernstein, M.D.1,3
1University of Manitoba Inflammatory Bowel Disease Clinical and Research Centre, Winnipeg, Manitoba,
Canada; and Departments of2Clinical Health Psychology and3Internal Medicine, University of Manitoba,
Winnipeg, Manitoba, Canada
This study reports cross-sectional medication adherence data from year 1 of the Manitoba
Inflammatory Bowel Disease (IBD) Cohort Study, a longitudinal, population-based study of multiple
determinants of health outcomes in IBD in those diagnosed within 7 yr.
METHODS: A total of 326 participants completed a validated multi-item self-report measure of adherence,
which assesses a range of adherence behaviors. Demographic, clinical, and psycho-social
characteristics were also assessed by survey. Adherence was initially considered as a continuous
variable and then categorized as high or low adherence for logistic regression analysis to determine
predictors of adherence behavior.
RESULTS: Using the cutoff score of 20/25 on the Medication Adherence Report Scale, high adherence was
reported by 73% of men and 63% of women. For men, predictors of low adherence included
diagnosis (UC: OR 4.42, 95% CI 1.66–11.75) and employment status (employed: OR 11.27, 95% CI
2.05–62.08). For women, predictors of low adherence included younger age (under 30 versus over
50 OR 3.64, 95% CI 1.41–9.43; under 30 vs. 40–49 yr: OR 2.62, 95% CI 1.07–6.42). High scores
on the Obstacles to Medication Use Scale strongly related to low adherence for both men (OR 4.05,
95% CI 1.40–11.70) and women (OR 3.89, 95% CI 1.90–7.99). 5-ASA use (oral or rectal) was not
related to adherence. For women, immunosuppressant use versus no use was associated with high
adherence (OR 4.49, 95% CI 1.58–12.76). Low trait agreeableness was associated with low
adherence (OR 2.03, 95% CI 1.12–3.66).
CONCLUSIONS: Approximately one-third of IBD patients were low adherers. Predictors of adherence differed
markedly between genders, although obstacles such as medication cost were relevant for both men
(Am J Gastroenterol 2007;102:1417–1426)
Medication represents a cornerstone of modern treatment
use can help patients to induce remission and maintain those
gains over time (1–4). Nevertheless, the medication can be
costly and difficult to take with unpleasant side effects, all of
which may result in less than optimal adherence to the treat-
ment regimen and poorer outcomes for disease management.
In this context, treatment adherence has significant implica-
tions for patient well-being and treatment outcome. Given
that physicians typically have limited control over this facet
of treatment, adherence also represents the greatest point of
vulnerability in this intervention. This is particularly salient
in the treatment of chronic disease when patients may be
asked to adhere to a specific regimen for several years at a
this context when assessing hypertension and diabetes with
rates varying between 50% and 65% (5, 6). Hence, it is im-
IBD has previously been identified as a particularly high-
risk illness for poor adherence (7, 8). Individuals are often
many years. Treatments can be inconvenient and difficult to
follow. Furthermore, the disease has an unpredictable course
with potentially long periods of inactivity. IBD patients must
have a strong belief in their doctor’s plan and strong convic-
tions about the necessity of the treatment to optimally follow
through with the treatment plan (8). Nonadherence rates for
short-term therapy have been described as varying from 20%
to 40% and rise as high as 72% for longer-term therapies
1418 Ediger et al.
Low medication adherence for those with IBD has been
linked to male gender, younger age, a higher number of pre-
scribed medications, high disease activity, shorter disease
has highlighted the importance of doctor–patient relation-
ships, frequency of medical appointments, and beliefs about
medication in predicting medication adherence (8). Replica-
tion of these predictors across studies has been inconsistent,
however. As such, there is currently no well-validated profile
for IBD patients with low medication adherence. Many of
the studies done to date have been hampered by a variety of
methodological limitations including small and potentially
It has been challenging to develop valid and inexpensive
measures of adherence. Determination of high and low ad-
herence often differs based on methodology and the types
of medication under examination. Adherence has been mea-
macy data, assays of blood or urine, electronic medication
dispensers, and verbal reports of compliance (12). The more
objective methods of adherence have most often been used
in treatment studies where patients have frequent monitoring
and contact with treatment staff. This frequent monitoring
(designed to improve adherence) is clearly not characteristic
not typical. In situations where patients are receiving diverse
forms of treatment and have less frequent contact from treat-
ment staff, verbal report has been used most often because it
treatment recommendations. However, the degree of agree-
ment among these various forms of assessment for treatment
adherence can vary.
Standardized self-report approaches have proven to be an
efficient and effective method of determining medication ad-
herence (12). They have established validity, positively cor-
virological outcome (15). However, self-report on treatment
sures, and a limited scale of responses (yes/no). Commonly
arbitrary definition of adherence that is often dependent on
the answer to a single question.
An obstacle inherent in adherence research is obtain-
ing data that are representative of the broad spectrum of
the patient population regarding their adherence behavior.
That is, studies typically rely on patient samples seeking
treatment in specialty clinics (7, 8, 17). These data, while
important, may reflect a sampling bias by assessing pa-
tients in specialized follow-up settings. Patients may be
more likely to have active disease and/or more positive be-
liefs about the medical system under these circumstances.
Our aim was to study factors related to treatment adher-
ence in a population-based community sample of those with
IBD that includes patients across the spectrum of disease
MATERIALS AND METHODS
ing on participants from the University of Manitoba IBD Re-
at least 18 yr of age and diagnosed within the previous 7 yr.
idents of the province of Manitoba, Canada (population ap-
administrative health database of Manitoba Health (the gov-
to all residents), were eligible for inclusion in the registry. Of
those eligible, that is, all those with IBD in the province, just
over half participated in the Registry (18).
versity of Manitoba Health Research Ethics Board and par-
ment in the research. At the time of study recruitment, there
were 3,192 participants in the Research Registry, of which
606 were eligible for this study, given the age and recent on-
set criteria. A total of 492 could be contacted over a period
of 18 months, and of those, 418 agreed to enroll in the study.
Complete data were obtained in the first contact of the longi-
they have subsequently served as the cohort. More details on
our group (19). Complete data on medication adherence and
and were collected 12 months after entry into the cohort.
The mean age of participants was 41 yr (SD = 14.06),
with a range from 18 to 80 yr. Sixty percent of partici-
pants were female. Forty-eight percent of participants were
taking 5-ASA, 21% were taking immunosuppresants, and
5% were taking prednisone. The average duration of disease
was 5.4 yr (SD = 2.1). The sample was 95% Caucasian,
with few having self-described backgrounds as East Indian,
Hispanic, or Metis. Approximately two-thirds of patients
were categorized as married (65%) and employed full time
(66%). One quarter of the sample was university educated
Assessment of Disease Type and Activity
IBD diagnosis subtype was verified through chart review in
the early stages of cohort development. A total of 162 partic-
ipants in this substudy (50%) had Crohn’s disease (CD), 146
(45%) had ulcerative colitis (UC, either ulcerative colitis or
ulcerative proctitis), and 18 (5%) had indeterminate colitis.
Individuals with indeterminate colitis were excluded from
categorical logistic regressions. Information regarding dis-
measurement and was determined by patient report of symp-
response scale. Active disease was defined as experiencing
symptoms constantly to occasionally at some point during
Predictors of Medication Adherence in IBD 1419
the prior 6 months, and inactive disease was defined as expe-
riencing infrequent symptoms or feeling well. Standardized
CD (20) and Powell-Tuck for UC (21) obtained during the
clinical interview, were used to validate self-reported disease
activity and these findings are described in a previous study
(19). Active disease in the previous 6 months was reported
by 66% of respondents, whereas 34% reported they had no
disease activity in that same period.
Assessment of Medication Adherence and Related
herence Report Scale (MARS) (22–25), which measures a
range of adherence behaviors. Nonadherence in this context
is operationalized by both the tendency to avoid, forget, or
stop taking medication, and the tendency to adjust or alter
the dose from that recommended by the physician (26). The
continuum of adherence behavior, in contrast to other scales
with dichotomous responses. Items are aggregated so that
no single item determines categorization as adherent. The
MARS allows patients to describe their typical pattern of
has been found to have adequate reliability, as well as good
criterion and discriminative validity (22, 27). The MARS
has been used to assess medication adherence in a variety of
health populations, including asthma, COPD, chronic pain,
published data validating this measure with an IBD sample
(Horne, personal communication, 2/18/06).
A few versions of this scale exist. While all nine items
were collected in this study (see Table 1), the most recent
five-item version (MARS-5) was used for all statistical anal-
yses. Items are scored on a 5-point Likert scale and added
to create a cumulative score with a total of 25 indicating
complete adherence. The scale can be analyzed either con-
Table 1. Adherence Behaviors from the Medication Adherence Re-
port Scale and Proportion of Respondents Indicating Frequent Dif-
ficulties in These Areas
% Adherence Items
I alter the dose∗
I forget to use it∗
I stop taking it for a while∗
I only use it if I feel sick
I decide to miss out on a dose∗
I take less than instructed∗
I avoid using it if I can
I use it only if I have to, if other things don’t work
I use it regularly every day
N = 326;∗Items on the five-item version of the Medication Adherence Report Scale.
Rating descriptions for this scale were: never, rarely, sometimes, often, and very often.
tinuously or categorically. Based on the positive skewing of
the sample distribution and consultation with the developer
of the MARS (Horne, personal communication, 2/18/2006),
those with scores of 20 or higher were categorized as hav-
ing high adherence. This cutoff is close to the mean and also
represents 80% of a total adherence score. A total of 213 par-
ticipants (65%) were classified as having high adherence and
113 (35%) as having low adherence.
Several other adherence-related measures were also col-
tionnaire (BMQ) (32) is a 17-item standardized scale assess-
ing specific concerns about medication the person is taking
vide a total score, with a higher score indicating greater con-
cern about medications. Psychometric data suggest that this
ulations including those being treated for asthma, diabetes,
cardiac problems, and psychiatric illness (32).
In addition, two self-report scales were developed specif-
ically for this study to assess potential obstacles and use of
both scales were generated based on the general adherence
literature, as well as clinical experience and patient report.
The Obstacles to Medication Use Scale is a list of 10 discrete
items concerning difficulties in taking medication regularly
(see Table 2 for the items on this scale). Each item had a
5-point Likert response scale; the items were summed and a
higher score indicated a greater number of perceived barri-
ers. The scale was found to have good internal consistency
used a 5-point Likert response scale with 7 discrete items
Table 2. Items on the Obstacles to Medication Use Scale and Pro-
portion of Respondents Indicating Frequent Difficulties in These
% Types of Obstacles
I forget to take the medication when it is
I am too busy to take the medication
The medication causes unpleasant side effects
The medication is costly
I feel ill at times when I am scheduled to take
I don’t have the medication with me when I
am scheduled to take it
Taking the medication reminds me that I have
had problems with IBD
I am not sure that the medication makes me
There are too many pills to take all at once
I have to take the medication too often during
N = 326.
1420 Ediger et al.
listing different methods people use to remind themselves to
take medication regularly. The responses were summed, and
a higher score equates to more use of reminders. It was antic-
ipated that greater medication concerns and obstacles would
be associated with lower adherence, whereas greater use of
reminders would be associated with higher adherence.
Assessment of Psychological Variables
Standardized self-report and interview measures were used
to assess psychological functioning. All of these measures
are widely used with established satisfactory psychometric
properties. Distress was measured using the Brief Symptom
Inventory (BSI) (33), a 53-item scale that assesses common
mental health symptoms. The Health Anxiety Questionnaire
(HAQ) is a 21-item scale measuring extent of health con-
cerns and somatic focus (34). The Neuroticism Extraversion
Openness–Five Factor Inventory (NEO-FFI) (35) is a sta-
ble personality inventory that assesses the five core person-
ality factors identified in empirical research: Neuroticism,
Extraversion, Openness, Agreeableness, and Conscientious-
ness. The inventory has 60 items rated on a 5-point Likert
scale. The Mastery Scale (36, 37) was used to assess the per-
son’s perceived control or efficacy in their life. All of these
measures were collected in the same period as the adherence
on entry into the cohort 12 months earlier.
Descriptive analyses were carried out for the demographic,
adherence, and psychological variables. The distribution of
der, socioeconomic status, distress, health anxiety, beliefs
about medication, mastery, and personality variables were
examined, and participant’s scores on continuous variables
were categorized as high or low based on the observed dis-
tribution of the sample to evaluate the relationships through
logistic regression. Subsequent analysis with logistic condi-
ings of gender differences in variables related to adherence
(9), the presence of a main effect of gender in our data (odds
a large number of gender interactions with other variables,
models were run separately for men and women to facilitate
interpretation of findings. In one model, however, small cell
associations between psychological variables and adherence
(with men and women combined due to small cell sizes),
cles to adherence. Minor variations in sample size among the
three analyses were caused by attrition due to small amounts
of missing data in the predictor variables. Both the second
and third model controlled for the effects of significant de-
mographic variables from the first model. ORs and the cor-
responding 95% CIs were estimated.
Adherence and Beliefs About Medication
Cumulative scores on the MARS-5 ranged from 8 to 25
with an average score of 20.67 (SD = 4.01). Responses
were skewed toward higher levels of adherence. Using the
score of 19 or lower to identify low adherence, approxi-
mately 35% of participants were classified as having low
adherence. Table 1 summarizes the frequency of adherence
behaviors. Eleven percent reported often or very often alter-
ing the dose or taking less than instructed. Likewise, 11%
also reported only taking their medication when they felt
sick. Interestingly, while forgetfulness has been reported to
play a key role in adherence difficulties (16), this item re-
ceived the lowest amount of endorsement with only 5% of
Responses to the Medication Reminders Scale suggested
that the most common memory strategy (often or very of-
ten used by 78% of respondents) was “using your own day-
time cues such as taking medication when brushing teeth,
with meals, or at bedtime.” Other strategies were much
less common with only 17% using a medication dispenser,
6% using a calendar or checklist, 6% a reminder from a
family member, 3% a blister pack provided by the phar-
macy, and 0% using a medication timer. The total score
on the Medication Reminders Scale had a relatively low
correlation with the total MARS-5 score (r = 0.13, P <
0.05) probably because few respondents used most types
of reminders, skewing responses to the low end of the
Item responses on the Obstacles to Medication Use Scale
are summarized in Table 2. Total scores were negatively cor-
related with responses on the MARS-5 (r = −0.37, P <
0.01) indicating that as the number of obstacles to medica-
tion use increased, adherence scores dropped. Among these
items, cost of medication was the most salient issue with
25% of patients indicating that cost often makes it difficult to
(13%), unpleasant associations with medication (13%), un-
certainties about the effectiveness of their medication (12%),
and forgetfulness (10%).
Participants’ perceptions about IBD medication were
largely positive, based on their responses to the Beliefs about
Medication Questionnaire. While 62% of respondents re-
ported some degree of worry about the long-term effects
of their medication, an equal number (62%) also believed
their medicines kept them from getting worse. A majority of
participants (76%) did not believe that their medication was
disruptive to their life. This perspective appears to have ex-
tion is addictive (61%) or that medications typically do more
harm than good (76%). In contrast, 45% of respondents en-
dorsed the idea that stopping medications periodically was a
Predictors of Medication Adherence in IBD1421
Table 3. Relationship Between Reported Adherence and Demographic and Disease Variables by Gender
Adherence Adherence Low Adherer
% Low OR for Being a% High
Adherence Adherence Low Adherer CI (95%)
% Low OR for Being
Demographic FactorsMF CI (95%)
Under 30-yr old
50-yr old and over
4 yr or less
5 yr or more
38 1.680.60–4.67 1.290.66–2.54
36 1.290.43–3.86 0.74 0.36–1.51
420.93 0.32–2.70 1.220.57–2.59
38 1251.620.55–4.74 0.95 0.49–1.85
∗P < 0.05;∗∗P < 0.01.
M = male; F = female; OR = odds ratio; CI = confidence interval.
OR >1 indicates more nonadherence; OR <1 indicates more adherence.
Male: Not working vs working (OR 11.27, 95% CI 2.05–62.08)
Female: Over 50 yr vs under 30 (OR 3.64, 95% CI 1.41–9.43)
Female: 40–49 yr vs under 30 yr (OR 2.62, 95% CI 1.07–6.42)
Demographic and Disease Variables Predicting Adherence
Logistic regression identified a few significant relationships
between medication adherence and both demographic and
disease variables (i.e., age, disease activity, diagnosis, mar-
ital status, employment, education, duration of disease) for
each gender. As can be seen in Table 3, the pattern of rela-
tionships differed between men and women. Men with lower
medication adherence were more likely to have a diagnosis
of UC (OR 4.42, 95% CI 1.66–11.75) and be employed on a
full-time basis (OR 11.27, 95% CI 2.05–62.08). These rela-
tionships were not significant for women, however. Age was
the only significant demographic predictor of medication ad-
herence in women. That is, women under 40 yr were more
likely to report low adherence than their older counterparts.
Age differences in adherence were not evident in men. Fur-
thermore, the proportion of men with low adherence across
the age range was generally similar to that of women over
40 yr of age. Disease activity, marital status, education, and
duration of disease were not significant.
Psychological Variables Predicting Adherence
Preliminary analyses of psychological predictors were con-
ducted examining men and women separately. However,
small cell sizes for several variables made this strategy un-
tenable. As a result, psychological predictors of adherence
were analyzed in a logistic regression with men and women
combined, controlling for gender, age, employment, diagno-
sis, and the interactions of gender with those variables. Once
these variables were accounted for, the overall model only
identified one additional variable that predicted medication
adherence. As shown in Table 4, a person’s level of agree-
ableness on the NEO-FFI significantly predicted adherence.
twice as likely to report low adherence (OR 2.03, 95% CI
1.12–3.66). While no other variables met statistical signifi-
cance, an examination of trends also suggests the utility of
distress as a potential predictor of adherence. This variable
marginally missed statistical significance and suggests the
possibility that more distressed individuals may also have
greater difficulty with medication adherence.
Medication Variables Predicting Adherence
ication characteristics helped to explain differences in ad-
herence. Type of medication and the frequency with which
that medication was administered were assessed through lo-
gistic regression controlling for age and employment. The
Obstacles to Medication Use Scale was also included in this
immunosuppressants. Data on prednisone were not included
1422Ediger et al.
Table 4. Relationship Between Reported Adherence and Psychological Variables
Low Logistic Regression
% Psychological VariablesN OR for Being a Low Adherer CI (95%)
Brief Symptom Inventory
Health Anxiety Questionnaire
Beliefs about medication
NEO-FFI (Personality Inventory)
Low openness to experience
High openness to experience
58 2.34 0.86–6.41
34 1.84 0.89–3.77
∗P < 0.05;∗∗P < 0.01 (N = 303).
Controlling for demographic factors: age, gender, employment, diagnosis, gender ∗ age, gender ∗ diagnosis gender ∗ employment.
OR = odds ratio; CI = confidence interval.
OR >1 indicates more nonadherence; OR <1 indicates more adherence.
High agreeableness vs low agreeableness (OR 2.03, 95% CI 1.12–3.66).
in this analysis because only 17 participants were taking this
medication at the time of the survey. Table 5 illustrates that
women were less likely to report low adherence when they
were taking immunosuppressants (OR 0.20, 95% CI 0.05–
0.77), or in other words, women on immunosuppressants
were more likely to report high adherence. This pattern did
not extend to women taking 5-ASA or to men. Obstacles to
3.94, 95% CI 1.92–8.08). There was no relationship between
treatment adherence and the number of pill administrations
per day. It is interesting to note that UC diagnosis remained a
significant predictor of low medication adherence (OR 5.89,
95% CI 2.06–16.88) for men even when these medication-
related variables were included in the analysis.
This study is the largest analysis of predictors of treatment
adherence for IBD and the first to use a population-based
sample. All participants were in their first 8 yr of illness.
generally consistent with adherence data from clinical trials
what occurs in usual management of patients with chronic
diseases (12). Predictors of adherence, on the other hand,
demonstrated a somewhat different pattern than that which
has been previously described.
Gender has been implicated as a significant predictor of
adherence in previous research (9), and was also found to be
relevant in our study. However, a markedly different pattern
of relationships was identified. Contrary to previous find-
ings, women in our sample were more likely to report low
adherence than men. Furthermore, gender had a clear effect
on the impact of other predictors. That is, all of the other
demographic predictors identified were only significant for
one gender or the other. Full-time employment and UC were
related to lower adherence, but only for men. Younger age
was predictive of more difficulties with adherence, but only
for women. Likewise, women taking immunosuppressants
reported better adherence than those on other medications,
variables were also significant during preliminary analyses
collapsed across gender. This highlights the importance of
Predictors of Medication Adherence in IBD 1423
Table 5. Relationship Between Reported Adherence, Medication, and Obstacles to Adherence
OR for Being a
OR for Being
Adherence Adherence Low Adherer
% Low% High
Adherence Adherence Low Adherer CI (95%)
5-ASA – no
5-ASA – yes
Immunosuppressants – no
Immunosuppressants – yes
Number of pill administrations
1–2 per day
3–4 per day
Obstacles to medication use
Low reported obstacles
High reported obstacles
MF CI (95%)
0.34 0.72–1.57 1.130.40–3.22
1.40 0.27–7.26 0.20∗∗
∗P < 0.05;∗∗P < 0.01.
Controlling for demographic factors: age, employment.
M = male; F = female; OR = odds ratio; CI = confidence interval.
OR >1 indicates more nonadherence; OR <1 indicates more adherence.
Immunosuppressants – yes vs immunosuppressants – no (OR 4.49, 95% CI 1.58–12.76).
looking for gender interactions in future adherence research.
Adherence has varied by gender in other studies and no sin-
gle profile of a low adherer specific to men or women has
In our study, for men, disease subtype was clearly relevant
for adherence. Those with Crohn’s disease were much more
Crohn’s disease may be more difficult to control. Certainly,
fistulas and bowel obstructions that may present in Crohn’s
disease and do not in UC may enhance patients’ desire to
adhere to a medication regimen. We have recently shown
that patients with Crohn’s disease have greater health-care
utilization relative to patients with UC (39). This is true for
both outpatient visits and hospitalizations. It is possible that
men with Crohn’s disease are more likely to be high adherers
either because of more severe disease requiring more health
care, or because of more health-care contacts that reinforce
It was anticipated that disease activity, disease duration,
and beliefs about medication would have an impact on treat-
ment adherence. Nevertheless, results of our study did not
levels of treatment adherence. This result is not surprising in
the light of previous findings regarding the importance of
the doctor–patient relationship and agreement for adherence
all was the measure of obstacles to adherence. The more
obstacles a person identified to adherence, the more likely
they were to have low adherence. This suggests that low
adherence is not merely carelessness or attitude, but a re-
sponse to realities in the patients’ lives. Financial pressures
and cost of medication, for instance, may force patients to
make difficult choices about when and how they will take
their medication. While it is possible that self-reported ob-
stacles are merely excuses created by poor adherers to ex-
plain their behavior, this is a testable hypothesis. Physicians
would only need to eliminate identified obstacles to see if
this changes adherence. These realities are important for
physicians to consider when they are planning a treatment
It is possible that adherence may be improved by engaging
the patient in creative problem solving or seeking treatment
Canada’s universal health-care program where basic medical
visits are covered and many individuals have good benefits
to cover much of the medication cost, cost was still the most
commonly reported obstacle to medication adherence. This
problem might be expected to be even more influential in sit-
and associated medication. Thus, a less costly drug may im-
for the patient. While not explicitly supported in this analy-
sis, previous studies have shown that a simpler regime (pills
taken once or twice a day vs multiple times in the day) can
improve adherence (40). Memory strategies such as medica-
may address unrealistic fears and associations the patient has
with their medication (41, 42). Engaging the patient in the
problem and having them help design a solution may be an
1424 Ediger et al.
empowering experience while at the same time highlighting
how important adherence is for their health.
While the broad definition of adherence as “doing what
the health professional recommends” (43) may be relatively
for adherence assessment has not yet been agreed upon. Fur-
ther, the level of adherence required for effective treatment
will vary depending on the medical condition and the type of
treatment. Self-report measures that have been standardized
and validated are well accepted in the adherence literature
This study contributes several factors not previously ad-
dressed in the adherence literature for IBD. The sample size
was the largest assessed to date to look at predictors of
treatment adherence, and participants were recruited from
a population-based sample, providing a broader spectrum
of IBD patients. While the fact that they have volunteered
for the project may suggest greater bias for adherence and
other socially conforming behaviors, this bias would not be
unique to this study. Further, this sample represents a closer
approximation of IBD patients in the community than pre-
viously available. As well, since individuals are not being
assessed by members of their treatment team, there may be
fewer of the demand characteristics that are frequently as-
sociated with adherence research. Finally, while there are
challenges in the measurement of adherence, the measure
used in this study assesses a range of adherence behaviors
and allows for differing degrees of adherence. It should be
noted that this measure assesses self-report of typical medi-
cation adherence—it does not directly assess adherence be-
havior in the context of a specific treatment as a pill count
This study considered many of the important variables re-
For example, there was no information available on the qual-
ity of the relationship with the prescribing physician or the
frequency of medical appointments. There were no measures
of psychiatric comorbidity, although the measure of distress
would capture some of this information. The primarily Cau-
casian sample should also encourage some caution in apply-
ing these findings to a more ethnically diverse population.
In addition to broadening the range of factors for concur-
rent assessment, future studies could incorporate more than
one method of assessing adherence, such as self-report and
administrative data that document prescription fills, to allow
an evaluation of agreement among different adherence mea-
In summary, we found that while many with IBD were
generally good at adhering to treatment, a significant portion
of individuals reported difficulties taking their medication
as directed. Thirty-five percent of participants met our cri-
terion cutoff for poor adherence behaviors. Our findings un-
derscore the complex relationships among adherence, demo-
graphic variables, psychological variables, and medication
variables. Of particular note were the differences between
men and women in these relationships. Self-reported obsta-
cles to adherence predicted low adherence in both men and
women. Brief screening of obstacles can promote problem
solving and dialogue between the patient and physician to
facilitate improved adherence.
This study was supported by a grant from the Canadian Insti-
tutes of Health Research. Dr. Bernstein is supported in part
entist Award. The authors would also like to thank Dr. R.
Horne of the University of London, U.K., for permission to
use the MARS and BMQ along with technical advice on the
What Is Current Knowledge
rThere is an increasing interest in understanding the
with inflammatory bowel disease (IBD).
rThere is currently no well-validated profile for IBD
patients with low medication adherence.
What Is New Here
rThis is the first population-based study of medication
adherence in IBD, and the largest sample size.
rApproximately one-third of IBD patients are low med-
rThere are different factors affecting medication adher-
ence in men versus women.
a diagnosis of ulcerative colitis (vs Crohn’s disease),
being employed, having high scores on the Obstacles
to Medication Use Scale and, having a low level of the
personality trait of agreeableness.
rFor women, important predictors of low adherence in-
Obstacles to Medication Use Scale, and having a low
suppressant use was associated with high adherence in
of medication was the most salient issue with 25% of
patients indicating that cost often makes it difficult to
take medication regularly.
Reprint requests and correspondence: Charles N. Bernstein,
M.D., 804F-715 McDermot Avenue, Winnipeg, Manitoba, Canada
Received December 19, 2006; accepted January 31, 2007.
Predictors of Medication Adherence in IBD 1425
Ann Rev Med 1995;46:497–505.
2. Kane S. Patient compliance and outcomes. Inflamm Bowel
3. Feagan BG, Fedorak RN, Irvine EJ. A comparison of
in Crohn’s disease. N Engl J Med 2000;342:1627–32.
4. Steinhart H. Maintenance therapy in Crohn’s disease. Can J
5. Nichol MB, Venturini F, Sung JCY. A critical evaluation of
Ann Pharmacother 1999;33:531–40.
prove medication-taking compliance. In:Burke LE, Ockene
LS, eds. Compliance in healthcare and research. New York:
Futura Publishing Co, 2001:57–70.
7. L´ opez San Rom´ an A, Bermejo F, Carrera E, et al. Adher-
Enferm Digest 2005;97:249–57.
8. Sewitch MJ, Abrahamowicz M, Barkun A, et al. Patient
Am J Gastroenterol 2003;98:1535–44.
herence with maintenance mesalamine in quiescent ulcera-
tive colitis. Am J Gastroenterol 2001;96:2929–33.
of noncompliance in inflammatory bowel disease. J Clin
11. Shale MJ, Riley A. Studies of compliance with delayed-
release mesalazine therapy in patients with inflammatory
bowel disease. Aliment Pharmacol Ther 2003;18:191–8.
12. Bosworth HB. Medication treatment adherence. In:
ment adherence, concepts, interventions, and measurement.
Mahwah, NJ: Lawrence Erlbaum Associates, 2006:147–
13. Haynes R, Taylor DW, Sackett DL. Can simple clinical
measurement detect patient noncompliance? Hypertension
14. Fleece L, Summers MA, Schnaper H, et al. Adherence to
pharmacotherapeutic regimen: Assessment and interven-
tion. Alabama J Med Sci 1988;25:133–8.
15. Haubrich RH, Little SJ, Currier JS, et al. The value
of patient-reported adherence to antiretroviral therapy in
predicting virologic and immunologic response. Califor-
nia Collaborative Treatment Group. AIDS 1999;13:1099–
16. Morisky DE, Green LW, Levine DM. Concurrent and pre-
dictive validity of a self-reported measure of medication
adherence. Med Care 1986;24:67–74.
17. Kane S, Huo D, Aikens J, et al. Medication nonadherence
and the outcomes of patients with quiescent ulcerative coli-
tis. Am J Med 2003;14:39–43.
18. Bernstein CN, Blanchard, JF, Rawsthorne P, et al. Epidemi-
ology of Crohn’s disease and ulcerative colitis in a central
Canadian province: A population-based study. Am J Epi-
19. Graff LA, Walker JR, Lix L, et al. The relationship of in-
functioning and quality of life. Clin Gastroenterol Hepatol
activity. Lancet 1980;1:514.
of oral prednisolone given as single or multiple daily doses
of Brighton, Brighton, UK, 2004.
23. Horne R. Non-adherence to medication: Causes and impli-
cations for care. In: Gard P, ed. A behavioural approach to
pharmacy practice. Oxford: Blackwell, 2001:111–30.
24. Horne R, Weinman J. Patients’ beliefs about prescribed
physical illness. J Psychosom Res 1999;47:555–67.
25. Horne R, Weinman J. Self-regulation and self-management
ment beliefs in explaining non-adherence to preventer med-
ication. Psychol Health 2002;17:17–32.
tool for audit and research. Qual Health Care 2001;10:135–
27. Ohm R, Aaronson L. The symptom perception and ad-
herence to asthma controller medications. J Nurs Scholar
medication nonadherence in patients with COPD. Chest
29. Kendrew P, Ward F, Buick D, et al. Satisfaction with infor-
mation and its relationship with adherence in patients with
chronic pain. Int J Pharm Pract 2001;9:R5.
to cholesterol-lowering medication in patients with hyper-
cholesterolaemia: The role of illness perceptions. Cardio-
vasc Drugs Ther 2004;18:475–81.
and European patients. N Z Med J 2004;117:10–8.
32. Horne R, Weinman J, Hankins M. The Beliefs about
Medicines Questionnaire: The development and evaluation
of a new method for assessing the cognitive representation
of medication. Psychol Health 1999;14:1–24.
scoring, and procedures manual, 3rd Ed. Minneapolis, MN:
National Computer Systems, Inc., 1993.
34. Lucock MP, Morley S. The Health Anxiety Questionnaire.
Br J Health Psychol 1996;1:137–50.
35. Costa PT Jr, McCrae RR. Revised NEO Personality Inven-
FFI) professional manual. Odessa, FL: Psychological As-
sessment Resources, 1992.
36. Pearlin LI, Schooler C. The structure of coping. J Health
Soc Behav 1978;19:2–21.
37. Pearlin LI, Menaghan EG, Lieberman MA, et al. The stress
process. J Health Soc Behav 1982;22:337–56.
38. Osterberg L, Blaschke T. Adherence to medication. N Engl
J Med 2005;353:487–97.
39. Longobardi T, Bernstein CN. Health care resource uti-
lization in IBD. Clin Gastroenterol Hepatol 2006;4:731–
40. Morningstar BA, Sketris IS, Kephart GC, et al. Variation
in pharmacy prescription refill adherence measures by type
of oral antihyperglycemic drug therapy in seniors in Nova
Scotia, Canada. J Clin Pharm Ther 2002;27:1–8.
41. Burgoon JK, Pfau M, Parrott R, et al. Relational communi-
pliance in communication between physicians and patients.
Commun Monogr 1987;54:307–24.
42. Roter Dl, Hall JA, Merisca R, et al. Effectiveness of inter-
ventions to improve patient compliance: A meta analysis.
Med Care 1998;36:1138–61.
1426Ediger et al. Download full-text
43. DiMatteo MR. Variations in patients’ adherence to medi-
cal recommendations: A quantitative review of 50 years of
research. Med Care 2004;42:200–9.
CONFLICT OF INTEREST
Guarantor of the article: Charles Bernstein
Specific author contributions: Jason Ediger: study plan-
ning, data analysis, writing of the paper, and final review of
John Walker: study planning, data analysis, writing of the
paper, final review of the paper.
Lesley Graff: study planning, data analysis, writing of the
paper, final review of the paper.
of the paper.
Ian Clara: statistical consultation, final review of the paper.
Trish Rawsthorne: study planning, data collection, final re-
view of the paper.
Linda Rogala: study planning, data collection, final review
of the paper.
Norine Miller: study planning, data collection, final review
of the paper.
Cory McPhail: data collection, final review of the paper.
Kathleen Deering: data collection, final review of the
Charles Bernstein: study planning, data analysis, writing of
the paper, final review of the paper.
Potential competing interests: None.