Hospital Pharmacy - July/August 2018 - 258

258
its use to the entire pharmacy staff to identify clinical and
technology issues before dissemination to the entire staff. An
electronic communication emphasizing the importance of
pharmacy interventions being documented in the electronic
medical record was also sent to the clinical pharmacy group.
After PCDST implementation, regular feedback was sought
from clinical pharmacists regarding their experience with the
tool and areas for improvement.

Statistical Analysis
Chi-square analysis was used for nominal data including the
primary endpoint, NCCN guideline adherence, and baseline
characteristics. An unpaired 2-sided t test was used for continuous data. Z tests were used for safety data. The MannWhitney U test was used for ordinal data. Microsoft Excel
2010 was used for statistical analysis. We anticipated requiring a total of 70 patients, or 35 in each group, to detect an
estimated difference of 30% between groups with a β of 0.80
and 2-sided α of 0.05 based on 33% attainment of analgesia
in the baseline preimplementation group.8

Hospital Pharmacy 53(4)

Study Endpoints
All study endpoints are summarized in Table 3. The primary
endpoint, attainment of analgesia at 24 hours from admission, was met in 10 of 30 (33.3%) in the preimplementation
group and 14 of 32 (43.8%) in the postimplementation
group. While a 10.5% absolute difference was found
between groups, statistical significance was not reached (P
= .78). The secondary endpoint of mean frequency of nursing pain assessments was higher in the postimplementation
group compared with the preimplementation group (12 vs
7.4, P < .001).

Safety Endpoints
Safety endpoints (summarized in Table 4) included respiratory rate less than 10 breaths per minute, oxygen saturation
<90%, and use of naloxone during the entire admission.
Overall, rates of adverse events were rare and not different
between groups.

Influence of Time of Admission
Results
Development of PCDST
The PCDST, a clinical decision support tool, was designed as
an application within the University of Chicago Medicine
information system (Epic Systems Corporation, Verona,
Wisconsin). The tool enables a systematic and transparent
approach to patient prioritization based on the display of the
most recent and maximum pain score, which can be seen in
Figure 1. Patients with a most recent pain score of 8 or
greater are displayed as a red colored box in the column;
those with a pain score of 4 to 7 are displayed as blue.
Patients with a most recent pain score of 5 or greater are
displayed in the box as well. Those with a pain score of
greater than or equal to 4 in the past 24 hours are displayed
with a red down arrow signifying that while the patient's
pain has recently improved, review of the patient's pain regimen by a pharmacist may be beneficial. The pharmacist may
document that the pain management of a patient has been
reviewed by placing a "check mark" to keep track of patient
pain review status within the EMR. A screenshot of the
patient list and PCDST is displayed in Figure 1.

Baseline Characteristics
The study population included 39 patients in the preimplementation group and 32 patients in the postimplementation
group. The demographic information for the 2 groups were
similar, with the exception of a higher incidence of metastatic disease and a statistically nonsignificant higher median
opioid dose in the preimplementation group (Table 2).

When time of patient admission was analyzed with respect to
pain outcomes, fewer patients who were admitted in the
1501-2300 timeframe (5 of 15) achieved analgesia at 24
hours when compared with those admitted during 0701-1500
and 2301-0700 (7 of 12 and 4 of 5, respectively) (Figure 2).

Discussion
Pain and safety outcomes were assessed in opioid-tolerant
adult hematology/oncology inpatients presenting with severe
pain prior to and following the implementation of a pharmacist-managed PCDST. A nonsignificant 10% difference in
achievement of analgesia was found (P = .78); however, the
study was initially powered to detect a 30% difference based
on a previous study at our institution and trials documenting
the influence of a clinical pharmacist on improving guideline
adherence. The feasibility of conducting a study to detect this
level of difference was determined not to be feasible based
on resources available, as an estimated 776 patients would
need to detect the 10% difference that was found. For this
reason, further expansion of the study populations was not
pursued.
Factors such as lack of pharmacist familiarity with and
use of the pain scoring tool may play a role in the outcomes.
Although the clinical pharmacy staff was trained on the use
of the tool, the study inclusion date in the postimplementation group was immediately after training due to the timeline
of the project. This may not have allowed enough time for
the pharmacy staff to become familiar with the tool. It is also
difficult to assess pharmacist adherence with the tool and
whether pharmacists are using it consistently.

Table of Contents for the Digital Edition of Hospital Pharmacy - July/August 2018