Non-response to Communication Technology Outreach
for Beta-agonist Overuse in a Pragmatic Randomized Trial
of Patients with Asthma
Marsha A. Raebel, PharmD
, Susan M. Shetterly, MS
, Glenn K. Goodrich, MS
Courtney B. Anderson, MPH
, Jo Ann Shoup, PhD
, Nicole Wagner, MPH
, and Bruce G. Bender,
Institute for Health Research, Kaiser Permanente Colorado, Denver, CO, USA;
Skaggs School of Pharmacy and Pharmaceutical Sciences,
University of Colorado, Aurora, CO, USA;
Division of Pediatric Behavioral Health, National Jewish Health, Denver, CO, USA.
KEY WORDS: asthma; outreach; beta-agonist; response; pragmatic trial;
communication technology; telephone; electronic.
J Gen Intern Med 33(6):809–11
© Society of General Internal Medicine 2018
Evidence suggests that communication technology applica-
tions can improve treatment adherence.
However, a recent
Cochrane Review concluded insufficient evidence exists to
determine the effects of automated communication on manag-
ing chronic conditions such as asthma.
Because not all pa-
tients are receptive to communication technology interven-
tions, defining factors associated with non-response to elec-
tronic outreach can inform tailoring future interventions to
increase effectiveness. As part of a pragmatic trial targeting
asthma patients with too frequent refills of inhaled beta-
the objective of this work was to de-
scribe and compare patients who did versus did not respond to
a communication technology outreach. A higher asthma med-
ication ratio (AMR), defined as the ratio of asthma controller
medications (numerator, e.g., inhaled corticosteroids) to total
controller medications plus inhaled beta-agonists (denomina-
tor), is associated with better asthma outcomes.
esized that patients who did not respond to outreach would
have a lower AMR than patients who did respond.
This work was conducted at Kaiser Permanente Colorado
(KPCO), an integrated healthcare system with approximately
600,000 members in the Denver-Boulder area. In KPCO usual
care, when a patient refills a beta-agonist more frequently than
every 60 days, an asthma care coordinator (ACC) is notified
through the electronic health record (EHR) for patient follow-
up. Because there are many reasons a patient could overfill
(e.g., extra inhaler for gym), knowledge of current asthma
symptoms is important, but unavailable to the ACC until after
s/he contacts the patient. The electronic outreach was designed
to determine whether the patient currently had symptoms to
guide further contact.
The study included members aged ≥ 18 diagnosed with
persistent asthma and without chronic obstructive pulmonary
disease. Patients were randomized 1:1 to receive a question
about symptoms by text/call (call if phone was not text-
enabled) or email if they overfilled from 2/9/2017 through
11/8/2017. Patients who responded they had symptoms were
contacted by an ACC; patients without symptoms were not
contacted. Patients who did not respond were contacted by an
ACC if the ACC had not contacted them in the last 90 days
and there was EHR evidence of a recent exacerbation (i.e.,
asthma-related corticosteroid burst, urgent care or emergency
department visit, or hospitalization).
We compared responders and non-responders on character-
istics, asthma exacerbations, and AMR the year prior to out-
reach. In univariable comparisons, Chi-square, Fisher’s exact,
or the Mann-Whitney test was used. We ran a multi-variable
model and selected variables that individually explained ≥ 5%
of the full model log-likelihood
for presentation in a smaller,
adjusted model estimating Relative Risks (RR) and 95% con-
fidence intervals (95% CI) of non-response. Statistical analy-
ses employed SAS version 9.4 (SAS Institute Inc., Cary, NC).
The KPCO Institutional Review Board approved this study.
The requirement for informed consent was waived.
Altogether, 420 of 4953 (8.5%) patients randomized to text
(N = 284)/call (N = 136) and 398 of 5046 (7.9%) patients
randomized to email overfilled a beta-agonist and received
outreach. Non-responders had a lower mean AMR (P = 0.019)
(Table 1). Other characteristics associated with non-response
included email outreach (P < 0.001), less than high school
education (P = 0.021), lower family income (P < 0.001), and
missed appointments (P = 0.008).
Published online March 12, 2018