Relationship between adherence to inhaled corticosteroids and poor outcomes among adults with asthma
Regular use of inhaled corticosteroids (ICSs) can improve asthma symptoms and prevent exacerbations. However, overall adherence is poor among patients with asthma. Objective To estimate the proportion of poor asthma-related outcomes attributable to ICS nonadherence. We retrospectively identified 405 adults age 18 to 50 years who had asthma and were members of a large health maintenance organization in southeast Michigan between January 1, 1999, and December 31, 2001. Adherence indices were calculated by using medical records and pharmacy claims. The main outcomes were the number of asthma-related outpatient visits, emergency department visits, and hospitalizations, as well as the frequency of oral steroid use. Overall adherence to ICS was approximately 50%. Adherence to ICS was significantly and negatively correlated with the number of emergency department visits (correlation coefficient [ R ] = -0.159), the number of fills of an oral steroid ( R = -0.179), and the total days' supply of oral steroid ( R = -0.154). After adjusting for potential confounders, including the prescribed amount of ICS, each 25% increase in the proportion of time without ICS medication resulted in a doubling of the rate of asthma-related hospitalization (relative rate, 2.01; 95% CI, 1.06-3.79). During the study period, there were 80 asthma-related hospitalizations; an estimated 32 hospitalizations would have occurred were there no gaps in medication use (60% reduction). Adherence to ICS is poor among adult patients with asthma and is correlated with several poor asthma-related outcomes. Less than perfect adherence to ICS appears to account for the majority of asthma-related hospitalizations.
Relationship between adherence to inhaled
corticosteroids and poor outcomes among
adults with asthma
L. Keoki Williams, MD, MPH,
Manel Pladevall, MD, MS,
Hugo Xi, MS,
Edward L. Peterson, PhD,
Christine Joseph, PhD,
Jennifer Elston Lafata, PhD,
Dennis R. Ownby, MD,
and Christine C. Johnson, PhD, MPH
Mich, and Augusta, Ga
Background: Regular use of inhaled corticosteroids (ICSs) can
improve asthma symptoms and prevent exacerbations.
However, overall adherence is poor among patients with
Objective: To estimate the proportion of poor asthma-related
outcomes attributable to ICS nonadherence.
Methods: We retrospectively identiﬁed 405 adults age 18 to 50
years who had asthma and were members of a large health
maintenance organization in southeast Michigan between
January 1, 1999, and December 31, 2001. Adherence indices
were calculated by using medical records and pharmacy claims.
The main outcomes were the number of asthma-related
outpatient visits, emergency department visits, and
hospitalizations, as well as the frequency of oral steroid use.
Results: Overall adherence to ICS was approximately 50%.
Adherence to ICS was signiﬁcantly and negatively correlated
with the number of emergency department visits (correlation
coefﬁcient [R]=20.159), the number of ﬁlls of an oral steroid
(R = 20.179), and the total days’ supply of oral steroid
(R = 20.154). After adjusting for potential confounders,
including the prescribed amount of ICS, each 25% increase in
the proportion of time without ICS medication resulted in
a doubling of the rate of asthma-related hospitalization
(relative rate, 2.01; 95% CI, 1.06-3.79). During the study
period, there were 80 asthma-related hospitalizations; an
estimated 32 hospitalizations would have occurred were there
no gaps in medication use (60% reduction).
Conclusions: Adherence to ICS is poor among adult patients
with asthma and is correlated with several poor asthma-related
outcomes. Less than perfect adherence to ICS appears to
account for the majority of asthma-related hospitalizations.
(J Allergy Clin Immunol 2004;114:1288-93.)
Key words: Adherence, asthma, attributable risk, hospitalizations,
inhaled corticosteroids, emergency department visits
In 2001, an estimated 31.1 million adults in the United
States had a diagnosis of asthma.
Asthma is a leading
cause of preventable hospitalizations, and it accounts for
an estimated 14 million days of missed school and 100
million days of restricted activity yearly.
It has been
estimated that regular use of inhaled corticosteroids (ICSs)
could reduce asthma hospitalizations by as much as
and that the risk of death from asthma decreases
by 21% for each additional ICS canister used.
such potential, patients with asthma appear to adhere
poorly to prescribed ICS medications. For example,
Cochrane et al
found that patients with asthma took their
ICS as directed on only 20% to 73% of days. On 24% to
69% of days, patients took <50% of the prescribed dose.
Although routine use of ICS can markedly improve
symptoms and reduce asthma complications,
tribution of medication nonadherence to poor asthma
control has not been established. Therefore, the objective
of this study was to measure the association between
adherence to ICS and the frequency of asthma-related
outpatient visits, emergency department (ED) visits,
hospitalizations, and oral steroid use.
This study was approved by the Institutional Review Board
and was in compliance with its Health Insurance Portability and
Accountability Act policy. Patients were all members of a large
health maintenance organization (HMO) in southeast Michigan, and
they received their care from a large, multispecialty medical group
consisting of approximately 800 physicians. Medical care claims data
the Department of Internal Medicine,
the Center for Health Services
the Department of Biostatistics and Research Epidemiology,
Henry Ford Health System, Detroit; and
the Section of Allergy and
Immunology, Department of Pediatrics, Medical College of Georgia.
Supported in part by grants from the Fund for Henry Ford Hospital, the
National Institute of Allergy and Infectious Diseases (R01AI50681), and the
National Heart, Lung, and Blood Institute (R01HL068971).
Received for publication August 4, 2004; revised September 22, 2004;
accepted for publication September 24, 2004.
Reprint requests: L. Keoki Williams, MD, MPH, Center for Health Services
Research, 1 Ford Place, 3A CHSR, Detroit, MI 48202. E-mail: kwillia5@
Ó 2004 American Academy of Allergy, Asthma and Immunology
BHR: Bronchial hyperreactivity
CMA: Continuous, multiple-interval measure of
CMG: Continuous, multiple-interval measure of
ED: Emergency department
HMO: Health maintenance organization
ICS: Inhaled corticosteroid
RR: Relative rate
Asthma diagnosis and
were used retrospectively to identify individuals with asthma.
Individuals age 18 to 50 years with 1 or more of the following in
the index year (January 1, 1999, to December 31, 1999) were
considered to have asthma: (1) 1 or more hospitalizations with the
primary discharge diagnosis of asthma, (2) 1 or more ED visits with
the primary diagnosis of asthma, or (3) 4 or more outpatient visits
with asthma as a diagnosis. These criteria were adapted from claims-
based measures of asthma that have been reported elsewhere.
examine the relationship between medication adherence and out-
comes among adult patients with asthma, data had to be available
for the 2 years after the index year. Therefore, to be included in the
analyses, patients with asthma had to be continuously enrolled in the
HMO between January 1, 1999, and December 31, 2001; have both
medical and pharmacy beneﬁt coverage during this time; and have
their care provided by physicians in the medical group. Patients were
excluded if they had a diagnosis of chronic obstructive pulmonary
disease or congestive heart failure at any outpatient visit, ED visit, or
hospitalization between January 1, 1999, and December 31, 2001.
The medical records of all individuals fulﬁlling these criteria
for asthma were abstracted for ICS use and dosage information.
Abstracted information was recorded and entered into an electronic
database by using TELEform (Cardiff Software, Vista, Calif) by
accredited medical record abstractors. These data were then linked
with prescription ﬁll information from pharmacy claims data. The
number of days that a given ﬁll of an ICS would last (ie, days
supplied) was calculated by dividing the canister size (ie, puffs per
canister) as derived from National Drug Codes in pharmacy claims
by the dosage information (ie, puffs per day) obtained in medical
records. The calculated days’ supply was used to estimate 2 measures
of adherence: (1) continuous, multiple-interval measure of medica-
tion availability (CMA), and (2) continuous, multiple-interval mea-
sure of medication gaps (CMG). The calculation of these indices is
discussed in detail by Steiner and Prochazka.
Brieﬂy, CMA is
cumulative days’ supply/total number of days between reﬁlls during
the observation period, and CMG is total days of treatment gaps/total
number of days between reﬁlls during the observation period.
Although complementary, CMG indices assess the effect of lapses in
therapy (increases with increasing gaps or delays in reﬁlls), whereas
CMA indices assess the effect of cumulative drug dosage (increases
with increasing frequency of prescription ﬁlls).
These 2 indices
were calculated for each patient beginning with the ﬁrst ﬁll of an ICS
and ending with the last ﬁll within the period of January 1, 2000,
through December 31, 2001. Therefore, ICS adherence could be
calculated only for individuals with at least 2 ﬁlls of an ICS in the
2-year observation period (n = 176). These calculations accounted
for changes in dosage and type of ICS. To ensure that gaps in ICS
reﬁlls were not a result of clinicians stopping medication, we also
abstracted provider notes for encounters in which an ICS was explicitly
stopped and another ICS was not started. Among patients for whom
adherence could be measured, we identiﬁed only 1 patient who had ICS
stopped while adherence was being measured and another ICS was not
started contemporaneously. Inclusion or exclusion of this patient did
not materially change the results; this individual was included in all
Adherence indices were calculated as previously described.
Subjects for whom adherence could be calculated were compared
with individuals for whom adherence could not be calculated.
Differences in characteristics were assessed by using a x
TABLE I. Characteristics of adults age 18 to 50 years with asthma by the availability of adherence information,
2000 to 2001
Characteristic Total (n = 405)
adherence information (n = 176)
adherence information (n = 229) P-value*
Age, y 6 SD 38.3 6 8.8 40.4 6 7.9 36.6 6 9.1 .001
Female (%) 270 (66.7) 119 (67.6) 151 (65.9) .751
African American (%) 192 (47.4) 75 (42.6) 117 (51.1)
White (%) 200 (49.4) 94 (53.4) 106 (46.3)
Other (%) 13 (3.2) 7 (4.0) 6 (2.6)
Medication use in 2000-2001
1 ICS ﬁll (%) 231 (57.0) 176 (100.0) 55 (24.0) .001
Mean ICS ﬁlls per person (6SD) 3.5 6 5.3 7.6 6 5.8 0.3 6 0.8 .001
1 inhaled b-agonist ﬁll (%) 281 (69.4) 168 (95.5) 113 (49.3) .001
Mean inhaled b-agonist ﬁlls per
7.4 6 9.6 13.5 6 10.8 2.6 6 4.7 .001
1 oral corticosteroid ﬁll (%) 183 (45.2) 118 (67.1) 65 (28.4) .001
Mean oral corticosteroid ﬁlls
per person (6SD)
1.5 6 2.9 2.6 6 3.7 0.7 6 1.8 .001
Medical encounters in 2000-2001
1 ED visit for asthma (%) 101 (24.9) 49 (27.8) 52 (22.7) .248
Mean ED visits for asthma
per person (6SD)
0.5 6 1.2 0.6 6 1.5 0.4 6 1.1 .151
1 hospitalization for asthma (%) 38 (9.4) 23 (13.1) 15 (6.6) .038
Mean hospitalizations for asthma
per person (6SD)
0.2 6 0.8 0.3 6 1.1 0.1 6 0.4 .022
4 outpatient visits for asthma (%) 145 (35.8) 104 (59.1) 41 (17.9) .001
Mean outpatient visits for asthma
per person (6SD)
3.8 6 5.2 6.2 6 6.3 2.0 6 3.2 .001
*For the comparison of those with adherence information to those without such information.
Age on January 1, 1999.
J ALLERGY CLIN IMMUNOL
VOLUME 114, NUMBER 6
Williams et al 1289
Asthma diagnosis and
categorical variables and a nonparametric Mann-Whitney 2-sample
test for continuous variables.
The primary outcome measures were the number of asthma-
related outpatient visits, asthma-related ED visits, asthma-related
hospitalizations, and oral steroid uses during the period of January 1,
2000, through December 31, 2001. The outcome measures were
related to CMA and CMG by using a Spearman correlation
coefﬁcient. Poisson regression methodology was then used to assess
the relationship between adherence indices and the frequency of
A univariable model was ﬁt, followed by
2 multivariable models. Multivariable models adjusted for potential
confounders including potential markers of disease severity, such as
the number of b-agonist ﬁlls,
the number of ICS ﬁlls, and the
prescribed number of ICS ﬁlls. The ﬁrst multivariable model (model
1) adjusted for sex, race, age, number of b-agonist ﬁlls, and number of
ICS ﬁlls (log-transformed). The second multivariable model (model
2) replaced the number of ICS ﬁlls with prescribed number of ICS
ﬁlls. This last variable was computed as the number of ICS ﬁlls di-
vided by CMA (log-transformed). The model was ﬁt with a correc-
tion for overdispersion. A relative rate (RR) was estimated in each
model for a 0.25 change in either CMA or CMG.
We adapted the methodology of Greenland and Drescher
calculate the proportion of hospitalizations attributable to non-
adherence. The coefﬁcient estimates from the multivariable
Poisson model (model 2) were used to derive an asthma-related
hospitalization rate for each individual with adherence information
(n = 176) under conditions of perfect ICS adherence (ie, CMG set to
0). This approach allowed us to account for other individual risk
factors that could contribute to the rate of hospitalization. The
expected number of asthma-related hospitalizations among persons
with adherence information was therefore the sum of these individual
rates. The number of asthma-related hospitalizations attributable to
nonadherence was then calculated by subtracting the expected
number of asthma-related hospitalizations from the observed number
of hospitalizations among patients with adherence information.
The proportion of all asthma-related hospitalizations attributable to
nonadherence was estimated by dividing this difference by the total
number of asthma-related hospitalizations observed among the 405
adults studied (ie, persons with and without adherence information).
This calculation assumes that the only hospitalizations that could
have been prevented by improved adherence were among the 176
subjects for whom adherence was measured. The asthma-related
hospitalizations that occurred in the 229 (56%) patients without
adherence information were considered immutable and therefore
contributed only to the calculation of the denominator, the total
asthma-related hospitalizations, but not to the numerator, the number
of asthma-related hospitalizations preventable by improved adher-
ence. Statistical analyses were performed by using SAS v8.0 (SAS
Institute, Cary, NC).
In the year 1999, 2150 patients between the ages of 18
and 50 years met the study criteria for having asthma. Of
these adult patients with asthma, 461 were continuously
enrolled in the HMO; had both medical and pharmacy
beneﬁts; and had their care provided by a group physi-
cian between January 1, 1999, and December 31, 2001.
Another 56 patients were excluded because they had
a diagnosis of chronic obstructive pulmonary disease or
congestive heart failure. The characteristics of the remain-
ing 405 patients are shown in Table I. Approximately
equal numbers were African American (n = 192) and
white (n = 200). Two-hundred thirty-one (57.0%) patients
had at least 1 ICS ﬁll in either 2000 or 2001, whereas 281
(69.4%) had at least 1 prescription ﬁll of an inhaled
b-agonist. The adherence measures, CMA and CMG,
could be calculated for 176 patients (ie, patients who had
at least 2 ﬁlls of an ICS in the years 2000 and 2001). Not
surprisingly, those for whom adherence could be mea-
sured had greater numbers of ICS ﬁlls compared with
those for whom adherence could not be measured. The
former also had other indicators of more severe asthma,
such as greater use of b-agonist inhalers, greater use of
oral corticosteroids, and more frequent asthma-related
hospitalizations. The mean CMA and CMG for ICS were
0.50 (SD, 60.37) and 0.54 (SD, 60.27), respectively. In
other words, overall adherence to ICS, as estimated by
CMA or 1-CMG, was approximately 50%.
As expected, CMA, a measure that increases when
patients ﬁll their ICS prescriptions on time, was negatively
associated with poor asthma outcomes (Table II). Sim-
ilarly, CMG, which increases with increasing lapses or
delays in ﬁlling prescriptions, was positively associated
with poor asthma outcomes. Both CMA and CMG were
signiﬁcantly correlated with the number of asthma-related
ED visits and the frequency of oral steroid use (both total
number of prescriptions ﬁlled and total number of days
treated). The weakest associations were seen for the
frequency of outpatient asthma visits, which did not
appear to be correlated with level of adherence. Slightly
stronger correlations were seen between adherence and
asthma outcomes when using CMG compared with CMA.
Because the relationships seen between adherence
indices and asthma outcomes could be reﬂective of
total ICS use rather than adherence alone, we simulta-
neously adjusted for inhaled steroid use, age, sex, race,
and b-agonist in the regression models ( Table III). In
TABLE II. Correlation between measures of adherence
to ICSs and asthma-related outcomes in adult patients
with asthma, 2000 to 2001
Variable 1 Variable 2 Correlation
CMA Asthma-related outpatient visits 20.015
Asthma-related ED visits 20.159*
Asthma-related hospitalizations 20.130
Fills of oral steroids 20.179*
Days of oral steroids 20.154*
CMG Asthma-related outpatient visits 0.030
Asthma-related ED visits 0.171*
Asthma-related hospitalizations 0.147
Fills of oral steroids 0.190*
Days of oral steroids 0.164*
CMA denotes continuous, multiple interval of medication availability
which is equal to the cumulative days’ supply of inhaled steroids divided
by the total number of days between reﬁlls for the period January 1, 2000
through December 31, 2001.
CMG denotes continuous, multiple-interval measures of medication gaps
which, on the basis of days’ supply of inhaled steroids, is equal to the total
number of days of without an inhaled steroid divided by the total days
between reﬁlls for the period January 1, 2000 through December 31, 2001.
J ALLERGY CLIN IMMUNOL
1290 Williams et al
Asthma diagnosis and
model 1, we adjusted for the total number of ICS ﬁlls made
by individual patients in 2000 and 2001. Because the pre-
scribed ICS dose may be a proxy for disease severity, in
model 2, we adjusted by the total number of ICS ﬁlls that
an individual should have had in 2000 and 2001 (ie, total
number of ICS ﬁlls divided by CMA). As shown pre-
viously, nonadherence as measured by increasing CMG
was positively, albeit not always signiﬁcantly, associated
with the rate of asthma-related ED visits, asthma-related
hospitalizations, and oral steroid use in both the un-
adjusted and adjusted regression analyses. In the un-
adjusted model and in model 2, CMG was signiﬁcantly
associated with the rate of asthma-related hospitalizations.
Each 25% increase in the proportion of time without ICS
medication resulted in a doubling of the rate of asthma-
related hospitalization (model 2, RR, 2.01; 95% CI, 1.06-
3.79). CMG was also signiﬁcantly associated with the rate
of ﬁlls of an oral steroid (model 1, RR, 1.49; 95% CI, 1.10-
2.02). Simultaneously adjusting models 1 and 2 for other
potential markers of disease severity, such as oral steroid
use and asthma-related hospitalizations, in 1999 did not
substantively change our results (data not shown). CMA
was similarly signiﬁcantly associated with the rate of ﬁlls
of an oral steroid (model 1, RR, 0.75; 95% CI, 0.58-0.97),
meaning that each 25% increase in adherence to ICS as
measured by CMA was associated in a 25% reduction in
the rate of oral steroid ﬁlls (data not shown).
Between 2000 and 2001, there were 80 asthma-related
hospitalizations among the 405 adults with asthma. Fifty-
eight of asthma-related hospitalizations occurred among
the 179 patients for whom adherence could be measured,
and 22 hospitalizations occurred in those without adher-
ence information. Among those for whom adherence could
be measured, we estimate that the number of hospitaliza-
tions would have been reduced to 10.0 (82.8% reduction)
were there no lapses in medication use (ie, CMG = 0). In
other words, 60.0% of all asthma-related hospitalizations
could be attributed to less than perfect adherence to ICS.
A previous study by Donahue et al
found that patients
with asthma who ﬁlled a prescription for an ICS were 50%
less likely to be hospitalized for asthma compared with
patients who did not receive an inhaled steroid (relative
risk, 0.5; 95% CI, 0.4-0.6). However, the investigators did
not examine whether patients took their ICS as directed.
We ﬁnd that apart from the total number of ICS ﬁlls, gaps
or lapses in adherence are an independent predictor of
asthma-related hospitalizations. In fact, we estimate that
60% of all asthma-related hospitalizations in our study
population could be attributed to less than perfect
adherence to ICSs.
These ﬁndings comport with trials showing that
cessation of ICSs can result in rapid declines in FEV
morning and evening peak expiratory ﬂow, and quality
of life, as well as increases in bronchial reactivity, daily
asthma symptom scores, sputum eosinophils, and levels of
exhaled nitric oxide.
For example, Haahtela et al
showed that average FEV
and bronchial hyper-reactivity
(BHR) to methacholine did not change signiﬁcantly in
individuals switched from budesonide at 1200 mg per day
to 400 mg per day, whereas switching to placebo resulted
in a signiﬁcant drop in FEV
and a rise in BHR. Vathenen
found that after 6 weeks of therapy with budeso-
nide, BHR returned to baseline within a week of
Although several studies have shown that overall
patient adherence to ICS is poor,
few have examined
the relationship between adherence and asthma outcomes.
In a small study by Milgrom et al,
adherence to ICS was
measured by using electronically monitored metered-dose
inhalers in 24 asthmatic children age 8 to 12 years. During
this 13-week study, children who required a burst dose of
oral corticosteroids had a median ICS adherence value of
13.7%, compared with a median ICS adherence of 68.2%
in children not requiring oral steroids (P = .008). In the
current study, we have also found a signiﬁcant correlation
between the need for oral steroids and adherence to ICS.
We did not, however, ﬁnd a signiﬁcant relationship be-
tween adherence and outpatient visits for asthma, which is
not surprising, because the latter may not reﬂect poor
This study must be interpreted in light of its limitations.
First, the study population examined here consisted
of individuals with both medical and pharmaceutical
coverage. On the basis of studies showing a positive
TABLE III. Unadjusted and adjusted association between adherence and asthma-related outcomes in adults
age 18 to 50 years, 2000 to 2001
Per 25% increase in CMG
Outcome Unadjusted RR (95% CI) Model 1* RR (95% CI) Model 2y RR (95% CI)
Outpatient visit 1.07 (0.93-1.23) 1.13 (0.97-1.30) 0.96 (0.83-1.11)
Asthma-related ED visit 1.36 (0.97-1.92) 1.27 (0.85-1.77) 1.25 (0.84-1.85)
Asthma-related hospitalization 1.80 (1.04-3.14)à 1.27 (0.75-2.14) 2.01 (1.06-3.79)à
Fills of oral steroids 1.22 (0.99-1.50) 1.49 (1.10-2.02)à 1.26 (0.95-1.67)
Days of oral steroids 1.18 (0.88-1.59) 1.53 (1.00-2.36) 1.27 (0.85-1.88)
*Adjusted for sex, race, age, number of b-agonist ﬁlls from 2000 to 2001, and number of ICS ﬁlls from 2000 to 2001.
Adjusted for sex, race, age, number of b-agonist ﬁlls from 2000 to 2001, and prescribed number of ICS ﬁlls from 2000 to 2001 (ie, total number of ICS ﬁlls
divided by CMA).
J ALLERGY CLIN IMMUNOL
VOLUME 114, NUMBER 6
Williams et al 1291
Asthma diagnosis and
relationship between both commercial insurance posses-
sion and household income and adherence,
mates of overall adherence are likely to be greater than
those of the general population. Similarly, our measure-
ment of adherence was limited to persons with 2 or more
ﬁlls of an ICS during the study period, and therefore, these
estimates did not account for individuals who never ﬁlled
their ICS prescription or ﬁlled it only once. Watts et al
performed a study of primary nonadherence (ie, the
nonﬁlling of prescriptions) in Port Lincoln, South
Australia. Of the 359 asthma medication prescriptions
written, 108 (30%) were never ﬁlled in 6 months of
follow-up. Together, these data suggest that the proportion
of poor asthma outcomes attributable to nonadherence
may be greater than reported here. When estimating the
proportion of hospitalizations attributable to nonadher-
ence, we adjusted for other risk factors that could have
contributed to the rate of hospitalization. However, it is
also possible that persons who were nonadherent differed
from persons who were adherent in other unmeasured
ways, which could have accounted for some of the
difference in these rates. This may have resulted in an
overestimation of the proportion of hospitalizations
attributable to nonadherence. Before generalizing these
results, it is also important to note that our deﬁnition of
asthma may have identiﬁed patients with more severe
asthma who were more likely to beneﬁt from continuous
ICS use. Because pharmacy claims data were used to
measure adherence, we could not assess daily patterns of
ICS use. Therefore, although CMG, a measure of gaps in
ICS reﬁlls, was the better predictor of poor asthma
outcomes, we could not distinguish protracted period off
therapy from chronic or intermittent underdosing. This
prevented us from being able to correlate speciﬁc patterns
and timing of use with adverse asthma outcomes. Last,
depending on the statistical method used, nonadherence
to ICS was signiﬁcantly associated with different poor
asthma outcomes. However, the consistency of these asso-
ciations in both magnitude and direction across outcomes
suggests that statistical differences were a result of limited
power rather than a true lack of association. Despite this
limitation, nonadherence was still signiﬁcantly associated
with oral steroid use and asthma-related hospitalizations
after adjusting for multiple potential confounders.
Although this study demonstrates the important re-
lationship between ICS adherence and asthma-related
outcomes, improving both is likely to be difﬁcult. Bender
recently reviewed interventions to improve adher-
ence among patients with asthma. The authors found
that few behavioral or educational interventions were
very effective at improving adherence and asthma control,
despite considerable time investment in some cases.
Similar disappointing results have been seen for adherence
interventions in other disease conditions.
However, recent data suggest that adherence can be
improved when patients know that adherence is being
measured and this information is regularly discussed with
them. Onyirimba et al
measured adherence to ICS by
using electronically monitored metered-dose inhalers in
19 patients with asthma over a 10-week period. Patients
who received regular adherence feedback from their
clinician had sustained levels of adherence >70% com-
pared with those in the control group whose adherence rate
fell below 30% by week 10. In a study by Reddel et al,
patients’overalladherence to peak expiratory ﬂowmeasure-
ment was 89% over a period of 72 weeks when participants
were aware that peak expiratory ﬂow meter use was being
electronically recorded and these results were discussed
with them regularly.
Our ﬁnding that ICS adherence, as measured by using
both medical records and pharmacy claims data, is associ-
ated with important asthma outcomes suggests that these
measures may be clinically useful. However, although
medical records and pharmacy claims data are ubiquitous,
patient adherence information is not, and to date, there are
few studies looking at its introduction in clinical prac-
Developing a mechanism to feed back adherence
information routinely to both clinicians and patients may
result in improved adherence and better asthma control.
Our demonstration that ICS adherence can be estimated by
using prescription reﬁll information, and that these
measures are independently associated with important
asthma outcomes, is a ﬁrst step in that direction.
We thank Lula Adams for her coordination of this project, as well
as Cheryl Spoutz, Rita Montague, and all of the medical record
abstractors whose help and hard work made this study possible.
1. Self-reported asthma prevalence and control among adults—United
States, 2001. MMWR Morb Mortal Wkly Rep 2003;52:381-4.
2. Blais L, Suissa S, Boivin JF, Ernst P. First treatment with inhaled
corticosteroids and the prevention of admissions to hospital for asthma.
3. Suissa S, Ernst P, Benayoun S, Baltzan M, Cai B. Low-dose inhaled
corticosteroids and the prevention of death from asthma. N Engl J Med
4. Cochrane MG, Bala MV, Downs KE, Mauskopf J, Ben Joseph RH.
Inhaled corticosteroids for asthma therapy: patient compliance, devices,
and inhalation technique. Chest 2000;1 17:542-50.
5. Haahtela T, Jarvinen M, Kava T, Kiviranta K, Koskinen S, Leht onen K,
et al. Comparison of a beta 2-agonist, terbutaline, with an inhaled cortico-
steroid, budesonide, in newly detected asthma. N Engl J Med 1991;325:
6. Kerstjens HA, Brand PL, Hughes MD, Robinson NJ, Postma DS, Sluiter
HJ, et al. A comparison of bronchodilator therapy with or without
inhaled corticosteroid therapy for obstructive airways disease. Dutch
Chronic Non-Speciﬁc Lung Disease Study Group. N Engl J Med 1992;
7. Pauwels RA, Pedersen S, Busse WW, Tan WC, Chen YZ, Ohlsson SV,
et al. Early intervention with budesonide in mild persistent asthma:
a randomised, double-blind trial. Lancet 2003;361:1071-6.
8. Donahue JG, Weiss ST, Livingston JM, Goetsch MA, Greineder DK,
Platt R. Inhaled steroids and the risk of hospitalization for asthma. JAMA
9. (Reference withdrawn)
10. Turk A. Overview of HEDIS 2000 Asthma Measurement. Am J Manag
11. Steiner JF, Prochazka AV. The assessment of reﬁll compliance using
pharmacy records: methods, validity, and applications. J Clin Epidemiol
12. Choo PW, Rand CS, Inui TS, Lee ML, Cain E, Cordeiro-Breault M, et al.
Validation of patient reports, automated pharmacy records, and pill
J ALLERGY CLIN IMMUNOL
1292 Williams et al
Asthma diagnosis and
counts with electronic monito ring of adherence to antihypertensive
therapy. Med Care 1999;37:846-57.
13. Le CT. Applied categorical data analysis. New York: Wiley; 1998.
14. Greenland S, Drescher K. Maximum likelihood estimation of the
attributable fraction from logistic models. Biometrics 1993;49:865-72.
15. SAS Institute. SAS/STAT users guide. Version 8. Cary, NC: SAS
16. Haahtela T, Jarvinen M, Kava T, Kiviranta K, Koskinen S, Lehtonen K,
et al. Effects of reducing or discontinuing inhaled budesonide in patients
with mild asthma. N Engl J Med 1994;331:700-5.
17. Lazarus SC, Boushey HA, Fahy JV, Chinchilli VM, Lemanske RF Jr,
Sorkness CA, et al. Long-acting beta2-agonist monotherapy vs continued
therapy with inhaled corticosteroids in patients with persistent asthma:
a randomized controlled trial. JAMA 2001;285:2583-93.
18. Vathenen AS, Knox AJ, Wisniewski A, Tattersﬁeld AE. Time course of
change in bronchial reactivity with an inhaled cortico steroid in asthma.
Am Rev Respir Dis 1991;143:1317-21.
19. Apter AJ, Boston RC, George M, Norﬂeet AL, Tenhave T, Coyne JC,
et al. Modiﬁable barriers to adherence to inhaled steroids among adults
with asthma: it’s not just black and white. J Allergy Clin Immunol 2003;
20. Milgrom H, Bender B, Ackerson L, Bowry P, Smith B, Rand C.
Noncompliance and treatment failure in children with asthma. J Allergy
Clin Immunol 1996;98:1051-7.
21. Bender B, Wamboldt FS, O’Connor SL, Rand C, Szeﬂer S, Milgrom H,
et al. Measurement of children’s asthma medication adherence by self
report, mother report, canister weight, and Doser CT. Ann Allergy
Asthma Immunol 2000;85:416-21.
22. Lafata JE, Xi H, Divine G. Risk factors for emergency department use
among children with asthma using primary care in a managed care
environment. Ambul Pediatr 2002;2:268-75.
23. Apter AJ, Reisine ST, Afﬂeck G, Barrows E, ZuWallack RL. Adherence
with twice-daily dosing of inhaled steroids: socioeconomic and health-
belief differences. Am J Respir Crit Care Med 1998;157:1810-7.
24. Watts RW, McLennan G, Bassham I, el-Saadi O. Do patients with
asthma ﬁll their prescriptions? a primary compliance study. Aust Fam
Physician 1997;26(suppl 1):S4-6.
25. Bender B, Milgrom H, Apter A. Adherence intervention research: what
have we learned and what do we do next? J Allergy Clin Immunol 2003;
26. McDonald HP, Garg AX, Haynes RB. Interventions to enhance patient
adherence to medication prescriptions: scientiﬁc review. JAMA 2002;
27. Onyirimba F, Apter A, Reisine S, Litt M, McCusker C, Connors M,
et al. Direct clinician-to-patient feedback discussion of inhaled steroid
use: its effect on adherence. Ann Allergy Asthma Immunol 2003;90:
28. Reddel HK, Toelle BG, Marks GB, Ware SI, Jenkins CR, Woolcock AJ.
Analysis of adherence to peak ﬂow monitoring when recording of data is
electronic. BMJ 2002;324:146-7.
29. Bieszk N, Patel R, Heaberlin A, Wlasuk K, Zarowitz B. Detection of
medication nonadherence through review of pharmacy claims data. Am J
Health Syst Pharm 2003;60:360-6.
30. Schectman JM, Schorling JB, Nadkarni MM, Voss JD. Can prescription
reﬁll feedback to physicians improve patient adherence? Am J Med Sci
J ALLERGY CLIN IMMUNOL
VOLUME 114, NUMBER 6
Williams et al 1293
Asthma diagnosis and