Keywords Best practice alert - Clostridium difficile - hospital-acquired infection
Introduction
Clostridium difficile infection (CDI) is a leading cause of hospital-associated gastrointestinal
illness with associated significant impact on patients' morbidity and mortality. In
the United States (US), CDI was estimated to be responsible for almost half a million
infections and associated with approximately 29,000 deaths in 2011.[1 ] It places a high burden on the US health-care system and hospitals length of stay.
Centers for Medicare and Medicaid Services (CMS) requires all positive CDI tests be
reported to the National Healthcare Safety Network (NHSN). CDI is adopted by CMS as
a new outcome measure in the hospital value-based purchasing program, which adjusts
CMS hospitals' payments effective January 2017.[2 ]
NHSN uses LabID Event reporting system, which is a much less labor-intensive standardized
method, based almost exclusively on laboratory data and limited admission date data,
including patient care location without clinical evaluation of the patient. It provides
proxy infection measures of C. difficile health-care acquisition, exposure burden,
and infection burden. Although this system is not highly specific to capture true
active infection versus colonization and does not identify the location where the
infection was truly acquired, it is still a practical simplified useful proxy for
CDIs.
For NHSN reporting purposes, CDI diagnosis is established with any CDI-positive laboratory
assay (reported as an event) irrespective of the signs, symptoms, and clinical presentation.[3 ] Categorization is based strictly on the number of hospital days between the specimen
collection date and the date the patient is admitted to the facility irrespective
of the clinical information
The crucial clinical and financial impact of CDI on patients, hospitals, and health
system resources has driven clinician, hospitals, and public health officials to collaborate
to reduce the burden of CDIs and to monitor and correctly identify the onset of CDIs.
Intervention
Our health system uses EPIC electronic medical records (EMR) across all its facilities.
It is an integrated health-care software suite which combines chart review, order
management, documentation for inpatient and outpatient care, and many other systems
for laboratory professionals, pharmacists, radiologists, and physicians. Using EMR
data, high-risk patients for CDI are identified on admission if they have two or more
of the following risk factors:
Previous history of CDI
Admission to a health-care facility in the previous 90 days
Antibiotic use during the previous 90 days.
Although there are many risk factors for C. difficile in the literature, the above
risk factors were prioritized for inclusion by the infectious disease and gastroenterology
expert opinion serving on the designated committee. Automated clinical decision support
tool was developed in EMR to display best practice alerts (BPAs) for nursing staff
and providers aiming to identify high-risk patients in the targeted timeframe. Both
BPA protocols were successfully piloted in a central facility and then expanded to
all Mercy facilities nationwide. Verbal teaching sessions and online education assignments
were done for the nursing staff to raise awareness about CDI impact and be familiar
with the protocol. All the BPAs went live on Monday, July 13, 2016, in the entire
Mercy Health System.
Nursing CDI protocol was developed to be applied during hospital days 1–3. When a
high-risk patient is identified on admission, the nurse will receive BPA-1 querying
if the patient has had 3 or more unformed stools in the previous 24 h. If the condition
is met, the nurse is prompted to collect a stool for CDI testing and start enteric
contact isolation ordered directly from the BPA alert. If not met, EMR settings will
continue to monitor stool documentations in the high-risk patients, and BPA-2 will
be triggered when three or more unformed stools have been documented within the first
three days of admission.
Providers CDI Protocol was developed to be displayed on the fourth day of admission.
When a high-risk patient has documented diarrhea without a laxative or bowel preparation
order, the provider will receive BPA-3 prompting to order CDI testing and enteric
contact isolation, otherwise, the reason for acknowledgment is required to dismiss
the alert. If a high-risk patient with no diarrhea is placed on antibiotics, the provider
will receive BPA-4 alert on the third day of the antibiotic course encouraging to
de-escalate antibiotics, order probiotics, and consider suppressive oral metronidazole
or oral vancomycin.
Methods
After obtaining approval from our Central Institutional Review Board, we have conducted
retrospective, and prospective cohort study in Mercy Hospital Fort Smith aimed to
evaluate the burden of CDIs in our facility and to determine whether BPAs can help
in the early detection of inpatient CDI and the reduction of inpatient HO-CDIs rates.
Pre- and post-BPA implementation periods were selected to be two years before and
after BPA release date on July 13, 2016, respectively, as shown in [Figure 1 ].
Figure 1: Schematic timeline and data collection flow for the best practice alerts
intervention study. CDI: Clostridium difficile infection, CO‑HCFA: Community‑onset
healthcare facility‑associated
Data collection
We have applied the LabID Event reporting methodology, definitions, and categorization
[Table 1 ]. Pre-BPA period data were collected including number of admissions, patient days,
and the overall facility-wide positive CDI events. Incident inpatients CDI cases were
identified. Duplicate results were removed. Incidence rates for CDI categories per
10,000 inpatient days and prevalence rates per 100 admissions were calculated per
CDC/NHSN Surveillance Definitions.[3 ] As NHSN reporting system includes pediatric cases, we have included them in our
analysis to be consistent with the standardized method of reporting. The same data
collection process was utilized for the first 6-month post-BPA period.
Table 1
Clostridium difficile infection key terms, categories, and definitions per National
Healthcare Safety Network LabID Event reporting system
Term
Definition
CO: Community-onset, CO-HCFA: Community-onset healthcare facility-associated, PCR:
Polymerase chain reaction, CDI: Clostridium difficile infection, HO: Healthcare-onset
CDI-positive
A positive laboratory test result for Clostridium
laboratory assay
difficile toxin A and/or B, (includes molecular assays (PCR) and/or toxin assays)
tested on an unformed stool specimen OR
A toxin-producing Clostridium difficile organism detected by culture or other laboratory
means performed on an unformed stool sample (must conform to the container)
Duplicate C.
Any Clostridium difficile toxin-positive laboratory
difficile-positive test
result from the same patient and location, following a previous Clostridium difficile
toxin-positive laboratory result within the past 2 weeks (14 days). There should be
14 days with no Clostridium difficile toxin-positive
Incident CDI LabID
Any CDI LabID event from a specimen obtained
event
>56 days (8 weeks) after the most recent CDI LabID event (or with no previous CDI
LabID event documented) for that patient
Recurrent CDI
Any CDI LabID event from a specimen obtained
LabID event Categorization
>14 days (2 weeks) and <56 days (8 weeks) after the most recent CDI LabID event for
that patient
CO
LabID event collected in an outpatient location or an inpatient location <3 days after
admission to the facility (i.e., days 1,2, or 3 of admission)
Healthcare
LabID event collected >3 days after admission to
facility-onset (HO)
the facility (i.e., on or after day 4)
CO-HCFA
CO LabID event collected from a patient who was discharged from the facility <4 weeks
before the current date of stool specimen collection. Data from outpatient locations
(e.g., outpatient encounters) are not included in this definition
Statistical analysis
In our interim analysis, two-sample tests of proportions were conducted to compare
pre- and post-BPA data using contingency tables. Pearson χ2-test was used to compare
proportions for appropriate cell count tables. Fisher's exact test was used for small
table cell size (n< 5). P < 0.05 is considered statistically significant. All calculations
were made using STATA version 14 (StataCorp, College Station, TX, USA).
Results
For 24 months before BPA implementation, 327 inpatient incident CDI cases out of total
428 facility-wide cases were detected (199 CO [60.3%], 88 HO [27.2%], 25 [7.8%] recurrent
and 15 CO-HCFA [5%]) among 30,424 admission and 104,990 patient days. Interim analysis
for the first 6-month period after the BPA implementation, 130 inpatient incident
CDI cases out of total 174 facility-wide cases were detected (98 CO [74.8%], 28 HO
[21.4%], and 5 CO-HCFA [3.8%]) among 9345 admissions and 24,346 patient days [Table 2 ].
Table 2
Data comparison before and after intervention
Average 6 months’ preintervention
6 months’ Postintervention
P
CDI: Clostridium difficile infection, CO: Community-onset, HO: Healthcare-onset, CO-HCFA:
Community-onset healthcare facility-associated
Denominator data
Patient days (n)
26248
24346
<0.001
Admission (n)
7606
9345
Numerator data
All positive CDI (n)
107
174
0.011
Outpatient, n (%)
23.25 (21.7)
38 (22.2)
Inpatient incident, n (%)
75.5 (70.6)
131 (75)
Inpatient recurrent, n (%)
6.25 (7.8)
0
Other, n (%)
2 (1.9)
5 (2.8)
Inpatient incident cases
75.5
131
0.080
CO, n (%)
49.75 (60.3)
98 (74.8)
0.066
HO, n (%)
22 (27.2)
28 (21.4)
0.094
CO-HCFA, n (%)
3.75 (5)
5 (3.8)
0.600
Overall prevalence rate
1.32
1.86
0.001
Admission prevalence rate
0.70
1.10
<0.001
CDI rate
38.38
71.47
CDI incidence rate
28.76
53.81
<0.001
HO incidence rate
8.38
11.50
0.143
Facility CDI combined
9.81
13.55
0.104
incidence rate
Discussion
In our interim analysis, we have observed higher CDIs incidence and prevalence rates
in the early post-BPAs phase. One explanation for this rise could be improved detection
using BPA.
For further evaluation, an A3 problem-solving process is being developed by our multidisciplinary
teams to give more insight after root cause analysis. Consideration for areas of improvement
includes compliance with hand hygiene, isolation precautions, antibiotics stewardship,
isolating high-risk patients on admission, and raising environmental services staff
awareness. The observed high CO-CDI proportion (almost 75%) is consistent with the
recent shift in CDI epidemiology over the last decade showing that CO-CDI rates are
dramatically increasing.[4 ] Raising public and patients awareness is a crucial step in CDIs control and prevention.
Our interim analysis has limitations. Failure to detect a significant change in the
HO-CDI proportion between the two groups may have been due to our small interim analysis
postintervention sample size. We were not able to measure the exact role of BPAs on
the observed higher incidence and prevalence as it could be related to nonmeasured
confounders in our study, such as antibiotics prescribing behavior, staff awareness
level about CDIs and their compliance with enteric isolation precautions before and
after the intervention. Nevertheless, our observation of a significant increase in
the overall incidence and prevalence rate may need further studies designed to measure
all the potential confounders, which is not our study preliminary outcome.
We have not included the patients' demographic and clinical characteristics in this
brief report as we are planning to present the big picture at the end of our study.
This study highlights the importance of EMR and BPAs utilization as useful tools to
enhance our understanding and management of diseases. Previous studies have utilized
EMR and BPAs to risk stratify HO-CDIs,[5 ] predict the risk for (HO-CDI) based on risk score [6 ] and to help with providing more standardized management.[7 ]
In summary, we conclude that short-term BPAs use is associated with significant increase
in the overall CDI infection rate, incidence rate, overall prevalence rate, and admission
prevalence rate along with a non-statistically significant decline in HO-CDI proportion.
Final analysis with full sample size is essential to provide a better picture about
the long-term effect.
