Customize your JAMA Network experience by selecting one or more topics from the list below.
Weinberger M, Murray MD, Marrero DG, et al. Effectiveness of Pharmacist Care for Patients With Reactive Airways Disease: A Randomized Controlled Trial. JAMA. 2002;288(13):1594–1602. doi:10.1001/jama.288.13.1594
Context It is not known whether patient outcomes are enhanced by effective pharmacist-patient
Objective To assess the effectiveness of a pharmaceutical care program for patients
with asthma or chronic obstructive pulmonary disease (COPD).
Design, Setting, and Participants Randomized controlled trial conducted at 36 community drugstores in
Indianapolis, Ind. We enrolled 1113 participants with active COPD or asthma
from July 1998 to December 1999. Outcomes were assessed in 947 (85.1%) participants
at 6 months and 898 (80.7%) at 12 months.
Interventions The pharmaceutical care program (n = 447) provided pharmacists with
recent patient-specific clinical data (peak expiratory flow rates [PEFRs],
emergency department [ED] visits, hospitalizations, and medication compliance),
training, customized patient educational materials, and resources to facilitate
program implementation. The PEFR monitoring control group (n = 363) received
a peak flow meter, instructions about its use, and monthly calls to elicit
PEFRs. However, PEFR data were not provided to the pharmacist. Patients in
the usual care group (n = 303) received neither peak flow meters nor instructions
in their use; during monthly telephone interviews, PEFR rates were not elicited.
Pharmacists in both control groups had a training session but received no
components of the pharmaceutical care intervention.
Main Outcome Measures Peak expiratory flow rates, breathing-related ED or hospital visits,
health-related quality of life (HRQOL), medication compliance, and patient
Results At 12 months, patients receiving pharmaceutical care had significantly
higher peak flow rates than the usual care group (P =
.02) but not than PEFR monitoring controls (P = .28).
There were no significant between-group differences in medication compliance
or HRQOL. Asthma patients receiving pharmaceutical care had significantly
more breathing-related ED or hospital visits than the usual care group (odds
ratio, 2.16; 95% confidence interval, 1.76-2.63; P<.001).
Patients receiving pharmaceutical care were more satisfied with their pharmacist
than the usual care group (P = .03) and the PEFR
monitoring group (P = .001) and were more satisfied
with their health care than the usual care group at 6 months only (P = .01). Despite ample opportunities to implement the program, pharmacists
accessed patient-specific data only about half of the time and documented
actions about half of the time that records were accessed.
Conclusions This pharmaceutical care program increased patients' PEFRs compared
with usual care but provided little benefit compared with peak flow monitoring
alone. Pharmaceutical care increased patient satisfaction but also increased
the amount of breathing-related medical care sought.
Reactive airways diseases, including asthma and chronic obstructive
pulmonary disease (COPD), are prevalent, morbid, and costly long-term conditions.1-8 Although
exacerbations are potentially preventable by appropriate drug therapy,9,10 patients often have difficulty following
prescribed regimens. Pharmacists may be able to enhance patients' compliance
and outcomes by engaging in pharmaceutical care activities (eg, monitoring
symptoms, providing medication counseling, helping resolve drug-related problems,
facilitating communication with physicians).11,12 Pharmaceutical
care is promising because: (1) pharmacists have the knowledge and skills to
identify and resolve patients' medication-related problems; (2) patients often
have several physicians but frequently patronize a single pharmacy; (3) pharmacists
are often the last health professional whom patients see before taking their
medication; and (4) pharmacists are trusted by patients.13
Literature reviews suggest that enthusiastic reports about the effectiveness
of pharmaceutical care are often plagued by serious methodological flaws.14-17 Although
inpatient pharmaceutical care may be effective,18,19 there
is less evidence of its effectiveness in outpatient settings and no rigorous
studies have been performed in community pharmacies.14-17 As
the largest provider of prescription services in the United States,20 community pharmacies provide an important venue for
improving patients' lives via pharmaceutical care. However, substantial barriers
to implementing such programs exist.21-23,25
We conducted a randomized controlled trial to assess the effectiveness
of pharmaceutical care for adults with reactive airways disease. Primary outcomes
were patients' peak expiratory flow rate (PEFR), health-related quality of
life (HRQOL), medication compliance, and breathing-related emergency department
(ED) or hospital visits. Secondary outcomes were patient satisfaction with
care and with their pharmacist.
The study, approved by the Indiana University-Purdue University at Indianapolis
institutional review board, was conducted in 36 Indianapolis CVS drugstores.
The 36 drugstores were divided into 12 clusters of 3 geographically proximal
drugstores ("triplets"). The 3 drugstores within each triplet were matched
on percentage of Medicaid-insured adults with reactive airways disease (to
control for customers' socioeconomic status) and number of prescriptions filled
(high vs low volume). Within each triplet, we used a random-number chart to
assign drugstores to 1 of 3 study groups. Each patient was followed up for
a year. Face-to-face interviews at baseline, 6, and 12 months were conducted
to assess primary and secondary outcomes (Figure 1).
Pharmaceutical Care Program. A detailed description of the pharmaceutical care program appears elsewhere.26 Briefly, the program included (Table 1) the following.
Pharmacist Training. Investigators representing several backgrounds presented: (1) an overview
of pharmaceutical care and its application to reactive airways disease; (2)
an orientation to the study computer and available patient-specific data;
(3) explanation for interpreting and using these data for pharmaceutical care;
(4) appropriate techniques for measuring PEFR; (5) study materials, resources,
and handouts when interacting with patients; and (6) strategies to implement
Computer Display of Patient-Specific Data. When a study patient filled any prescription (not only breathing medications),
the drugstore computer alerted pharmacists to review patient-specific data
contained in a separate study computer behind the counter. To safeguard patients'
confidentiality, access to patient-specific data required pharmacists' individualized
passwords. Study computers contained: (1) contact information for patients
and 1 to 2 physicians caring for their breathing problem; (2) graphical display
of all PEFR data gathered during monthly interviews; (3) dates and locations
of recent ED visits and hospitalizations; and (4) breathing medications (including
compliance rates and refill histories). These data were obtained during monthly
telephone interviews. Pharmacists were encouraged to document their pharmaceutical
care activities at the bottom of the screen.
Written Patient Educational Materials. We developed 1-page handouts corresponding to specific problems associated
with clinical data stored in the study computer. Handouts, designed to be
easily understood by patients, used mnemonic devices and color coding to facilitate
distribution by pharmacists.
Resource Guide. Attached to the study computer, guides contained laminated pages with
practical suggestions to help pharmacists implement the program in a busy
Pragmatic Strategies to Facilitate Pharmaceutical Care. To reinforce pharmacist training and facilitate program implementation,
we: (1) encouraged pharmacists to page the on-call investigator with questions;
(2) had an investigator make personal visits to each intervention drugstore
every 1 to 2 months; (3) distributed periodic newsletters containing information
about reactive airways disease and suggestions on implementing the program;
(4) faxed weekly lists of recent patient activity (eg, medication refill,
ED or hospital visit) and pharmacists' documented activities; and (5) provided
pharmacists with telephone appointment scheduling cards to facilitate interactions
with patients at a mutually convenient time. During the final year of the
study, we paid pharmacists $50 per month for high rates of compliance with
the pharmaceutical care protocol (viewing data on the study computer for ≥90%
of patients and documenting actions for ≥75% of patients).
Control Groups. Patients in the pharmaceutical care group received a peak flow meter,
instruction about its use, and monthly calls from research personnel to obtain
current PEFR results. We were concerned that these activities could be an
active treatment by increasing patients' self-monitoring. So, the peak flow
meter monitoring control group also received a peak flow meter, instructions
about its use, and monthly calls to elicit PEFRs. However, PEFR data were
not provided to the pharmacist. Patients in the usual-care group received
neither peak flow meters nor instructions in their use; during monthly telephone
interviews, PEFR rates were not elicited. Pharmacists in both control groups
also had a 4-hour training session although the topics were different and
they received no components of our pharmaceutical care program (Table 1).
Customers were eligible if they (1) filled a prescription for methylxanthines,
inhaled corticosteroids, inhaled or oral sympathomimetics, inhaled parasympathetic
antagonists, or inhaled cromolyn sodium during the preceding 4 months; (2)
reported having COPD or asthma as an active problem; (3) were 18 years or
older; (4) received 70% or more of their medications from a single study drugstore;
(5) reported no significant impairment in vision, hearing, or speech that
precluded participation; (6) did not reside in an institution (eg, nursing
home); and (7) provided written informed consent.
The recruitment protocol27 was designed
to maximize customers' confidentiality (Figure
1). Patients were enrolled between July 1998 and December 1999.
In July 1998, drugstore programmers queried their database to identify all
customers 18 years or older who, during the preceding 4 months, filled a prescription
for 1 of the above breathing medications at any study pharmacy. Corporate
personnel mailed letters to these individuals stating that: (1) they were
working with Indiana University School of Medicine (IUSM) to develop programs
to improve the health of its customers; (2) IUSM was evaluating these programs
by talking to customers; (3) customers would be paid up to $60 for participating
in the evaluation; and (4) they could page an investigator with questions
about the project. Subjects were asked to sign and return a form providing
permission in order to release their names to IUSM investigators or that they
were uninterested in participating. A drugstore employee attempted to telephone
persons who failed to return a signed form to verify that they received the
letter, determine their interest in participating, and offer to send them
another form. Only after receiving a signed form indicating a person's willingness
to be contacted was the name released to the IUSM project manager.
The project manager then conducted a telephone screening interview to
describe the study, review patient eligibility, and, for eligible patients,
arrange a face-to-face baseline interview (Figure 1). All interviewers completed training on the study protocol,
interviewing techniques, asthma and COPD, proper peak flow meter technique.
Interviewers were instructed to ensure that the proper peak flow meter technique
was used during face-to-face interviews.
Interviewers, blinded to study group assignment, obtained informed consent
and conducted baseline interviews. After completing an interview, the laptop
computer used to administer interviews revealed the patient's study group
assignment. At that time, interviewers distributed a Personal Best peak flow
meter (Health Scan Product, Inc, Cedar Grove, NJ) and reviewed proper meter
technique for patients in the pharmaceutical care program and peak flow monitoring
control groups. Following the baseline interview, we sent letters notifying
physicians treating the patient's breathing problem and advised them that
the patient was participating in a study in which a pharmacist might receive
PEFR, medication compliance, and health services utilization data. The letter
stressed that the pharmacist would make no treatment decisions but may educate
patients about their breathing problem and reinforce compliance with the physician's
prescribed treatment regimen.
Patients were censored from the study if they died, were placed in a
nursing home, moved away permanently from Indianapolis, their insurance no
longer covered using these drugstores, or they lost telephone access. For
patients not censored, we attempted to conduct in-person follow-up interviews
at 6 and 12 months to assess outcomes by individuals blinded to study group.
Patients in all 3 groups received a $20 gift certificate for each interview
Monthly telephone interviews were conducted with all patients to ascertain
ED or hospital use, breathing medications not contained in the pharmacists'
database (ie, over-the-counter, prescriptions from other drugstores, samples),
and frequency with which they had spoken to a pharmacist about their breathing
medications during the previous month. In addition, patients receiving peak
flow meters were asked to use their meters and report their PEFR while on
the telephone. Patients are capable of reliably recording PEFR readings.28-31
Peak expiratory flow rate, the maximum velocity of exhalation that can
be generated by patients, correlates highly with parameters of pulmonary function
and clinical outcomes of patients with asthma and COPD.32-36 For
this study, we transformed PEFRs into the percentage of maximum predicted
value based on patients' sex, age, and height.37
Disease-specific HRQOL was assessed during the baseline and 6- and 12-month
interviews using asthma- and COPD-specific measures developed to detect clinically
important changes within randomized trials.38-40 Both
questionnaires use analogous 7-point Likert formats to produce an overall
assessment, as well as disease-specific subscales for COPD (dyspnea, fatigue,
emotional function, mastery) and asthma (activity limitations, symptoms, emotional
function, environmental stimuli). Scores range from 1 (worst) to 7 (best)
function; differences of 0.5 units are considered clinically important. Patients
were administered the COPD questionnaire if they met American Thoracic Society
criteria (age >45 years and smoking history ≥10 pack-years).41,42 Otherwise,
patients completed the asthma questionnaire.
Medication compliance with breathing medications over the previous month
was assessed using 2 validated measures: a single-item indicator43 (proportion
of noncompliance), and a 4-item scale44 ranging
from 0 (low) to 4 (high) noncompliance. Self-report has been found to be valid
when inquiries are made in a nonthreatening manner.43-45
Breathing-related ED or hospital visits were obtained from patient reports
during monthly interviews and the Indianapolis Network for Patient Care, an
integrated network linking data from Indianapolis' major hospitals, neighborhood
health centers, and public health and homeless clinics.46 We
contacted the site of care to verify the visit. To determine whether an ED
or hospital visit was breathing-related, a physician and pharmacist, blinded
to patient's study group, independently reviewed reports for each episode.
Disagreements were adjudicated by the raters.
Patient satisfaction with care was assessed using a validated 4-item
global measure.47,48 Items were
modified to ask about satisfaction for their breathing-related problems. Scores
range from 4 (lowest) to 20 (highest) satisfaction. Patient satisfaction with
pharmacists was measured with an 11-item scale that has been used among patients
with reactive airways disease.49 Scores (average
of the 11 items) range from 1 (lowest) to 5 (highest) satisfaction.
A log file captured each time the intervention pharmacists had accessed
a patient's record from the study computer or had documented their actions
on the study computer. The frequency with which pharmacists documented their
actions was used to estimate the dose of the intervention.
All analyses were conducted with SAS Version 8 (SAS Institute, Cary,
NC). The success of randomization was assessed by comparing baseline characteristics
of study groups using χ2 tests for categorical variables and
analysis of variance for continuous variables. When assumptions for these
parametric tests were not met, we used the Fisher exact and Kruskal-Wallis
tests, respectively. When between-group baseline differences were significant
(P<.05), we controlled for those variables in
all subsequent analyses. We used χ2 tests and t tests to compare baseline characteristics of patients who did and
did not complete 12-month interviews. During the study period, 2 drugstores
(both control) were closed. In both cases, all study patients transferred
to another study drugstore. Patients' original study group was retained for
Consistent with our hypotheses, we compared the pharmaceutical care
group with each of the 2 control groups using an intent-to-treat analysis.
Sample size was based on our ability to detect a 10% difference in breathing-related
ED or hospital visits (25% vs 15%) between the pharmaceutical care group and
each control group with 80% power and α = .025 (using Bonferroni correction50 for 2 pairwise comparisons). For an estimated 15%
attrition, we enrolled approximately 1100 patients.
To test for differences in intervention group means for continuous outcomes
measured at 6 and 12 months, we used repeated measures analysis of variance
models, with baseline scores as covariates. This model makes no assumption
about outcome measures in intervention groups' being linear over time. Random
effects corresponding to the design variables (ie, triplet and pharmacy) were
included. However, because variation due to pharmacy was consistently nonsignificant,
we excluded the effect of pharmacy from the final models. For binary outcomes
(compliance, proportion with a breathing-related ED or hospital visit), we
used logistic regression. For compliance, we used a repeated measures approach.
For breathing-related ED or hospital visits, we weighted each person's contribution
by the amount of time in the study. Because HRQOL had different items depending
on disease group (asthma vs COPD), separate models were constructed. For all
other outcomes, interaction effects of disease with intervention group were
investigated. If the interaction was not significant, we analyzed COPD and
asthma patients together, including disease group as a covariate. Separate
models were analyzed for patients with COPD and for patients with asthma if
the interaction was significant. Treatment by time interactions were also
tested and, if significant, tests of intervention effect were done separately
for 6- and 12-month outcomes. We also examined temporal trends by comparing
6- and 12-month outcomes to baseline measures, accounting for multiple comparisons
using Holms procedures.51 If the overall effect
of intervention group was significant from tests described above, change from
baseline was analyzed separately for each intervention group. For all repeated
measures analyses, compound symmetry variance-covariance structure was used
to specify the correlation between responses for the same individual. For
all pairwise comparisons, we reported unadjusted P values.
As a secondary analysis, we used data from the pharmaceutical care program
group only to determine if there were a dose-response relationship with our
primary outcomes by repeating analyses with an additional continuous predictor
for dose: the number of times the pharmacist documented his/her actions on
the study computer.
Letters were mailed to 14 195 persons meeting initial eligibility
criteria: 3019 returned forms (2195 were interested; 824 refused). Of those
not returning a form, follow-up telephone calls identified 756 additional
persons interested in participating. Of 2951 interested patients screened
by the project coordinator, 1202 were ineligible, 492 refused, and 136 could
not be contacted. After excluding 8 pilot study participants, 1113 were enrolled. Table 2 presents baseline characteristics
for 453 patients with COPD and 660 patients with asthma. We completed interviews
with 947 patients (85.1%) at 6 months and patients 898 patients (80.7%) at
12 months. Completion rates did not differ significantly by disease or study
group. Patients not completing 12-month interviews were more likely to report
a hospital or ED visit during the month prior to enrollment (9.8% vs 5.9%, P = .04), less education (13.1 vs 13.6 years, P = .04) and, for COPD patients only, lower HRQOL. Unadjusted outcome
data across time are presented in Table
Study groups were comparable at baseline (P>.05),
except for race (both diseases) and PEFR (COPD only). To account for these
differences, we controlled for race in all analyses and for baseline PEFR
among COPD patients only. Except for satisfaction with health care, for which
the intervention group by time interaction was significant, intervention effects
were reported across 6 and 12 months together. Disease groups were tested
in the same model for all outcomes except breathing-related ED or hospital
visits, for which the effect of the intervention groups differed significantly
Across 6 and 12 months, there was a significant (P = .006) overall difference in PEFR among the 3 study groups (Table 4). Specifically, the pharmaceutical
care group had higher PEFR than usual care (P = .02)
but was not different from the peak flow meter monitoring group (P = .28). Time contrasts showed significant (P<.01)
increases in PEFR in the pharmaceutical care group from baseline to 12 months
and in the peak monitoring group from baseline to both 6 and 12 months.
There were no significant between-group differences in overall HRQOL
for patients with either asthma (P = .23) or COPD
(P = .31, Table
4). To increase statistical power, we combined patients with asthma
and with COPD, using overall HRQOL as the dependent variable and disease as
covariate; the treatment group difference remained nonsignificant (P = .12). For each disease group, we observed statistically significant
(P<.002) within-group improvement over 12 months
in overall HRQOL. This same pattern was observed for all HRQOL subscales in
Similarly, there were no significant between-group differences in medication
compliance using the proportion noncompliant (P =
.22) or 4-item scale (P = .57); there were significant
(P<.001) within-group declines in noncompliance
at 6 and 12 months.
There was a significant interaction (P = .02)
between disease and study group for the proportion of patients with a breathing-related
ED or hospital visit over 12 months. Although there was no difference (P = .34) across study groups for COPD patients (Table 4), asthma patients in the pharmaceutical
care group were more likely to have a breathing-related ED or hospital visit
than those receiving usual care (odds ratio [OR], 2.16; 95% confidence interval
[CI], 1.76-2.63; P<.001). There was no difference
between the pharmaceutical care and peak flow meter control groups (P = .32).
Across 6 and 12 months, pharmaceutical care group patients reported
greater satisfaction with their pharmacist than did patients in either the
usual care (P = .02) or peak flow monitoring (P = .001) control group. Moreover, the pharmaceutical care
group reported greater satisfaction with their health care compared with the
usual care group at 6 months (P = .01); this difference
was not sustained at 12 months (P = .14). There was
no difference in satisfaction with health care between the pharmaceutical
care and peak flow monitoring groups at 6 months (P =
.08) or 12 months (P = .14). During monthly interviews,
49.3% of patients in the pharmaceutical care group reported having discussions
with their pharmacists about their breathing problems while 24.3% of patients
in the peak flow meter monitoring and 24.2% in the usual care control groups
reported having discussions with their pharmacists (P<.001).
Patients in the pharmaceutical care group reported a mean (SD) rate of speaking
to their pharmacist during the previous month 1.17 (1.77) times compared with
0.51 (1.19) times in the peak flow monitoring and compared with 0.40 (0.89)
in the usual care group (P<.001). Frequency of
interactions with the pharmacist was correlated with patient satisfaction
with their pharmacist (r = 0.12; P<.001).
There were ample visits during which pharmacists had opportunities to
implement the pharmaceutical care program for patients with both asthma (mean
[SD], 19.4 [16.8] visits) and COPD (22.4 [17.7] visits). However, they only
accessed data from the study computer about half the time (asthma, 10.3 [7.5]
visits; COPD, 11.8 [10.5] visits) and documented actions only about half the
time these records were accessed (asthma, 6.2 [5.8] visits; COPD, 6.2 [7.0]
visits). Using documentation to assess dose of the intervention, we observed
no statistically significant dose-response effect for PEFR or HRQOL. Notably,
when pharmacists documented more pharmaceutical care actions, patients exhibited
less noncompliance with their breathing medications (OR, 0.96; 95% CI, 0.92-0.99; P = .02) and had more breathing-related ED or hospital
visits (OR, 1.06; 95% CI, 1.04-1.07; P<.001).
We examined the effectiveness of a pharmaceutical care program designed
to improve patients' clinical status, HRQOL, and medication compliance. At
12 months, patients in the pharmaceutical care group had significantly higher
PEFRs than those receiving usual care; however, our program offered no advantage
compared with monitoring patients' PEFRs monthly. Moreover, patients participating
in our program were significantly more satisfied with their pharmacists than
the other 2 groups; they were also more satisfied with their health care than
the usual care group. Although we observed a substantial improvement in both
medication compliance and HRQOL among patients in the pharmaceutical care
program at 6 months that was sustained at 1 year, similar improvement was
observed in both control groups.
However, contrary to our hypothesis, asthma patients in the pharmaceutical
care group had more breathing-related ED or hospital visits. Notably, we are
observing care-seeking behavior and did not examine the appropriateness of
visits although observed PEFRs suggest poor function. Interestingly, although
our a priori hypotheses did not involve comparing the 2 control groups with
each other, breathing-related ED or hospital visits in the peak flow monitoring
control group (14.6%) was twice that of usual care (7.3%). Increased visits
may have resulted from patients associating their PEFR values with symptoms,
which could have resulted in more care seeking. That simple monitoring could
have measurable effects on clinical outcomes is consistent with our previous
study of regular telephone calls from non–health care professionals.52 These findings suggest that strategies to increase
patient involvement in care of their chronic conditions (especially enhancing
self-efficacy for monitoring) may increase health services utilization.
The most likely explanation for our findings was that despite our efforts
to design a pragmatic program and reinforce its use, it was not used consistently.
Notably, pharmacists only viewed data in the study computer half the time
patients filled prescriptions, and they documented their actions only 50%
of the time those data were viewed. There are several possible explanations
for this. First, our intervention was cumbersome and required pharmacists
to access data on a separate study computer. When this investigation began,
it was not possible to integrate patient-specific data into the regular drugstore
computer. Now, those options exist through the intranet. Second, implementation
of the program may require release time or other incentives for the pharmacists.
Third, although all pharmacists participated in our program, they were not
universally enthusiastic about this expanded role. In retrospect, a better
strategy may have been to identify 1 enthusiastic pharmacist per store to
be responsible for implementing the program in his/her drugstore. Indeed,
a recent randomized trial found that an intervention delivered by highly motivated
and selected community pharmacists in Canada improved the process of cholesterol
risk management; interestingly, unlike our results, patients were not more
satisfied with this program.53
Our investigation had several limitations. First, although our overall
drop-out rate was relatively low (18%), those not completing 12-month interviews
appeared to have worse breathing problems, worse HRQOL scores, and less education.
These are characteristics of patients who might have benefitted most from
our program. Indeed, the rate of breathing-related ED or hospital visits,
particularly among patients with asthma, was lower than expected from our
previous work.54 Second, the recruitment protocol
required to safeguard patient confidentiality resulted in our inability to
contact more than half the patients, thus compromising generalizability. Third,
several important questions are beyond the scope of this study: Was the observed
increase in visits due to increases in ED or hospital use? Were the increased
visits clinically appropriate? Was there an effect on the appropriateness
of medication regimens? Could the increase in ED visits have accounted for
improvements in PEFR (eg, by alterations in medications)? Finally, we cannot
determine whether patients' knowledge that they were being observed influenced
Pharmaceutical care can play an important role in patient care, as supported
by a recent American College of Physicians position paper.55 However,
data from this study suggest that implementation of the pharmacy care program
was poor, perhaps due to limited time or lack of incentives to use the resources
provided, and resulted in limited benefits in terms of clinical end points.
Given the poor implementation, it is not surprising that our program provided
little benefit compared with peak flow monitoring alone. Additional research
is needed to determine whether other approaches using pharmaceutical care
will improve patients' outcomes. Such evaluations should be conducted in "real-world"
community pharmacies using strategies that are pragmatic in busy retail pharmacies,
even though randomized trials conducted in real world settings present methodological
and pragmatic challenges.
Create a personal account or sign in to: