Page 4 CirculationCardiovascularQualityandOutcomes

Magid et al Pharmacist-Led, Heart360 Home BP Program 3

patients, made medication adjustments as needed, and communicated For patients who purchased their medications from a KPCO phar-
with patients via telephone or secure e-mail. Any medication chang- macy, medication adherence was calculated from a medication pos-
es were communicated to the primary care physician of the patient session ratio based on the total number of days supplied for each
through the EHR. Patients who neglected to upload their BP read- filled antihypertensive medication, less the supply that would extend
ings as instructed received up to 3 reminder phone calls through an beyond the end of the 6-month study period, divided by the period for
automated interactive voice response system. If a patient still failed to which the medication was prescribed. For patients on multiple anti-
upload readings, he or she received a call from a clinic staff member. hypertensive medications during this time, adherence to each medica-
tion was averaged to derive a summary adherence measure.
Six-Month Visit
Patients in both groups returned for a clinic visit at 6 months, at Sample Size
which time they had their BP taken by a research assistant blinded This study was designed to enroll up to 200 patients per group al-
to study group assignment using the same standardized protocol that located equally to the HBPM and UC groups. Assuming a 15% drop-
was used at the baseline visit. In addition, all patients were asked to out rate and a control rate of 30% in the UC group, this sample size
rate their overall satisfaction with their hypertension care and the de- provided 80% power to detect a 14% difference in BP control rate in
gree to which they were engaged in their hypertension care during the the HBPM group compared with the UC group.
6-month study period. Patients in the HBPM group were also asked
about how easy it was to measure their BP at home and how easy it
was to upload their BPs to Heart360 and to rate their interactions with Statistical Analysis
the clinical pharmacy specialist. All statistical analyses were performed on an intention-to-treat ba-
sis with SAS version 9.1 software (SAS, Cary, NC). In the primary
Outcome Measures analyses, all patients randomized at baseline were included. Baseline
characteristics were reported as means, medians, and SDs for in-
The primary outcome was the proportion of patients who attained terval- and ratio-level variables (eg, age) and proportions for nomi-
their goal BP at the 6-month clinic visit. BP goals were <140/90 nal- and ordinal-level data (eg, sex, comorbidities). Interval-level
mm Hg for all patients except those with DM and CKD, whose outcome variables were assessed for normality of their distributions,
16
goal was <130/80 mm Hg. Secondary outcomes included change and appropriate tests were used to assess differences in mean values
in SBP and DBP between the baseline and 6-month clinic visits, between groups (eg, t test, rank-sum test). To assess differences in
change in antihypertensive medication intensity, and antihypertensive proportions between groups on categorical variables, the Pearson χ 2
medication adherence. Medication intensity was measured by test of association was used.
comparing the proportion of patients in each group with at least 1 There were 22 people who did not complete the 6-month follow-up
antihypertensive medication added between the baseline and the visit and were missing BP outcomes for this study. Two methods were
6-month visit and the proportion with at least 1 dose increase for an used to include all persons randomized at baseline in these analyses.
antihypertensive medication that they were taking at baseline. Medical For analyses of BP change, we estimated generalized linear models
service used, including all hospitalizations, emergency department with a separate record for each time period: baseline and 6-month
visits, clinic visits, telephone encounters, and e-mail encounters, was follow-up. Individuals missing outcome data at 6 months (n=22) have
assessed via chart review. Patients in the HBPM group were asked only a baseline record in this model, whereas all others contributed
to measure their home BP at least 3 times per week and to upload 2 records. The intervention effect was estimated via an interaction
readings weekly. Patients were considered to be adherent to the BP with time, assuming an unstructured covariance matrix and clustering
monitoring protocol if they measured and uploaded home BP readings within clinic estimated as a random effect. To help account for po-
for ≥80% of the weeks during the study intervention. The mean and tential differences in individuals missing 6-month follow-up data, the
median number of BP measurements per upload were also recorded. models included covariates for age, sex, race, number of medications,

Figure 1. Patient flow diagram. BP indicates
blood pressure; and HBPM, home blood pressure
monitoring.

Downloaded from circoutcomes.ahajournals.org by guest on March 20, 2013