Content uploaded by Maxim Topaz
Author content
All content in this area was uploaded by Maxim Topaz on Oct 15, 2018
Content may be subject to copyright.
Received: 23 July 2018
|
Accepted: 16 August 2018
DOI: 10.1002/nur.21907
RESEARCH ARTICLE
Improving patient prioritization during hospital-homecare
transition: A pilot study of a clinical decision support tool
Maxim
Q1
Topaz
1,2
|
MaryGrace Trifilio
3
|
Donna Maloney
3
|
Ofrit Bar-Bachar
1
|
Kathryn H. Bowles
3,4
1
Faculty of Social Welfare and Health
Sciences, School of Nursing, University of
Haifa, Haifa, Israel
2
Brigham and Women's Health Hospital,
Boston, Massachusetts
3
Visiting Nurse Service of New York, New
York, New York
4
School of Nursing, University of
Pennsylvania, Philadelphia, Pennsylvania
Correspondence
Maxim Topaz, Faculty of Social Welfare and
Health Sciences, School of Nursing, University
of Haifa, Haifa, Israel.
Email: mtopaz80@gmail.com
Funding information
Eugenie and Joseph Doyle Research
Partnership Fund
ABSTRACT
Patient admission to homecare is a complex process. Medicare policy requires that all
patients receive a first home visit within 48 hr after the referral is received at the
homecare agency. For unstable or high risk patients, waiting 48 hr to be seen by
homecare nurses may not be safe. In this pilot study we tested an innovative clinical
decision support tool (called PREVENT), designed to identify patients who may need to
be prioritized for early homecare visits. The study was conducted in 2016 at a large
homecare agency in the Northeastern US with 176 patients admitted to homecare
from the hospital. In the control phase (n= 90 patients), we calculated the PREVENT
priority score (indicative of high or medium/low first nursing visit priority) but did not
share the score with the homecare intake nurses who influence visit scheduling. In the
experimental phase, the PREVENT score was shared with the homecare intake nurses
(n= 86 patients). During the experimental phase, high-risk patients received their first
homecare nursing visit about one-half a day sooner than in the control phase (1.8 days
vs. 2.2 days). Rehospitalizations from homecare decreased by 9.4% between the
control (21.1%) and experimental phases (11.7%). This pilot study of patient
prioritization showed promising results: high priority patients received their first
homecare visit sooner and overall rehospitalization rates were lower.
KEYWORDS
clinical decision support, homecare, home health care, nursing informatics, patient
prioritization, transitional care
1
|
INTRODUCTION
Q2
Each year, homecare agency administrators and nurses working for the
approximately 12,000 homecare agencies across the United States (US)
admit more than 11 millionpatients to homecare services (The National
Association for HomeCare and Hospice, 2010). Registered nurses (RNs)
in administrative and clinical positions make several critical decisions
before and duringhomecare admission, including identifying patients at
risk for poor outcomes who might benefit from early interventions. Yet,
there are no empirically derived standards to assist in making these
important decisions. Nationwideevidence shows that about 12–20%of
those admitted to homecare services are rehospitalized during the
homecare episode (up to 60 days during which homecare services are
provided) (Centers for Medicare and Medicaid Services, 2016; MedPac,
2014). There is a growing body of evidence showing that a significant
portion of these rehospitalizations may be prevented by timely and
appropriately targeted allocation of healthcare services (Markley,
Sabharwal, Wang, Bigbee, & Whitmire, 2012;McDonald, King, Moodie,
& Feldman, 2008; Murtaugh et al., 2016; O’Connor, 2012). Patient
prioritization for homecare services might be key to better outcomes,
but there is a lack of decision support tools to help nurses prioritize
which patients might benefit from early nursing visits.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
Journal MSP No. Dispatch: August 29, 2018 CE:
NUR 21907 No. of Pages: 8 PE: Jennifer Chinworth
Res Nurs Health. 2018;1–8. wileyonlinelibrary.com/journal/nur © 2018 Wiley Periodicals, Inc.
|
1
2
|
THE PROBLEM: IDENTIFYING PATIENT
PRIORITY BEFORE THE FIRST HOMECARE
NURSING VISIT
The first homecare visit, usually conducted by a RN, is one of the most
important steps of the homecare episode for several reasons. This
start-of-care visit includes an examination of the home environment, a
discussion regarding whether a caregiver is present and able to help,
and an assessment of the patient's capacity for self-care. A unique care
plan is created based on this evaluation of the patient's needs. The RN
reconciles medications to avoid errors and assure adherence to the
treatment plan and reviews the hospital discharge instructions with the
patient and caregivers. These activities are important to prevent
adverse events that may lead to rehospitalization. For some patients,
the sooner this visit occurs, the better.
Currently, the Center for Medicare and Medicaid Services (CMS)
requires that all patients admitted to homecare services receive an
initial assessment within 48 hr of their referral to homecare (Centers
for Medicare & Medicaid Services, 2012). In our extensive literature
search, however, we did not identify empirical evidence to support this
regulation. According to the CMS, homecare agencies fail to see about
11% (SD 10.7%) of their patients within 48 hr of referral to homecare,
and there is a large variation of time to first visit among states/
territories (from 28.2% failure in the Virgin Islands to 5% in North
Dakota) (Centers for Medicare and Medicaid Services, 2016).
Homecare admission is a complex process that often involves
multiple care providers and healthcare organizations. For example,
some homecare agencies have hospital-based liaisons that help
coordinate and streamline homecare admissions, while other agencies
rely on hospital discharge planning staff or referring physicians to
discern the patient's condition prior to the first homecare visit.
Inefficient information flow can complicate the homecare admission
process and may result in delayed care and poorer patient outcomes
(Sockolow, Ms, & Bass, 2016).
Homecare agencies serve patients of varied clinical complexities,
from patients requiring just a few visits for assessment and instruction,
to patients with daily wound care or those who are ventilator-
dependent. The responsibility of patient prioritization and visit-timing
falls on the homecare agency's administrative nurses or their hospital-
based colleagues. They must decide whether patients need an
immediate visit or if the patients can wait. Accurate allocation for
the appropriate timing of the first visit is crucial, especially for patients
with urgent healthcare needs. In this pilot study, we focused on
patients admitted to homecare from hospitals rather than other
settings (e.g., skilled nursing facility).
3
|
WHY PATIENT PRIORITIZATION IS
NECESSARY AND IMPORTANT
As many as 60% of patients admitted to homecare from hospitals are
not ready for hospital discharge for a variety of reasons, including lack
of knowledge on how to deal with their complex medical conditions
(old and new) and poor self-management skills (Barnason, Zimmerman,
Nieveen, Schulz, & Young, 2012; Bobay, Jerofke, Weiss, & Yakusheva,
2010; Coffey & McCarthy, 2012; Foust, Vuckovic, & Henriquez, 2012;
Weiss, Yakusheva, & Bobay, 2010). Moreover, patients often do not
visit their primary care providers for follow-up visits within the first
60 days of hospital discharge, resulting in the homecare nurse
becoming the key clinician for their health needs (Grafft et al., 2010;
Hernandez et al., 2010; Jencks, Williams, & Coleman, 2009; Misky,
Wald, & Coleman, 2010; Sharma, Kuo, Freeman, Zhang, & Goodwin,
2010; Zisberg et al., 2011).
Typical homecare patients have at least four medical conditions
(Caffrey, Sengupta, Moss, Harris-Kojetin, & Valverde, 2011; Leung
et al., 2016) and take, on average, eight to twelve medications
(McDonald et al., 2012; McDonald et al., 2016). The combination of
lack of discharge readiness, poor follow-up with primary care
providers, and high medication complexity makes homecare patients
vulnerable to poorer outcomes. Operating with limited information,
resources, and time (Sockolow et al., 2016), administrative homecare
nurses and those nurses working in the home must decide which
patients to prioritize to prevent rehospitalizations and other negative
outcomes.
4
|
FIRST NURSING VISIT PRIORITIZATION
TOOL
Our team developed an innovative decision support tool called
PREVENT to assist administrative nurses in determining which
patients should be prioritized for the first homecare nursing visit
(Topaz, 2014). Tools similar to PREVENT have been developed
internationally. For example, researchers in Canada have developed a
tool called Method for Assigning Priority Levels (MAPLe)
Q3
(Hirdes
et al., 2008). This tool helps prioritize homecare patients for
community and other post-acute services. Tools like MAPLe, however,
were developed for different purposes and in different homecare
environments, possibly limiting their applicability in the unique US
healthcare system. To our knowledge, there are no other existing US-
specific homecare first-visit prioritization tools.
PREVENT was developed using data mining, regression
modeling, and expert homecare nurses’ratings of example patients
who were transitioned from hospital to homecare. The goal was to
identify key patient characteristics that are essential to support
early homecare admission decision making. Overall, more than 70
patient demographic and clinical characteristics (e.g., comorbid-
ities, level and availability of social support, detailed functional
status) were considered for inclusion in the final prediction model
from which PREVENT was developed. The total study sample
included 670 patients. Model development was done using 70% of
the sample for model development and 30% model validation was
done using the other 30% of the sample. The resulting prediction
model achieved an Area under the Curve (AUC) of 75.9 when
validated on the hold-out sample. Models with AUC > .70 are
considered valid (Hosmer et al., 2013).
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
2
|
TOPAZ ET AL.
In the PRVENT development study, nurses were asked to read
patients’discharge case summaries and identify key risk factors,
defined as any clinical or demographic factors that can help identify
high priority patients for the first homecare visit. The nurses chose the
following five patient risk factors as significantly associated with
prioritizing the patient's first homecare nursing visit: (a) presence of
wounds (either surgical or pressure ulcers); (b) a documented comorbid
condition of depression; (c) limitation in current toileting status
(requiring use of any assistive equipment, assistive person, or both); (d)
number of medications; and (e) number of comorbid conditions. These
factors, identified as significant predictors of the first homecare
nursing visit priority, were validated statistically (each factor was
statistically significant in the adjusted regression model) and confirmed
in the clinical literature on patient prioritization. For example,
depression, or multiple comorbid conditions are known as major
contributors to rehospitalizations from homecare (Ma, Shang, Miner,
Lennox, & Squires, 2017; O’Connor, 2012).
Each of the risk factors was assigned a specific score based on the
logistic regression weights from the original study (Topaz, 2014). For
example, if a wound was present (e.g., pressure ulcer, vascular ulcer),
the patient received a score of 15 points. For each additional co-
morbid condition, one point was added to the final score. The scores
for the factors were summed, generating a cumulative score ranging
between 0 and 50, with any score >26 suggesting high priority for the
first homecare visit. The cut-off of 26 points was established based on
the PREVENT development study regression model (Topaz, 2014).
More details on the methods are published elsewhere (Bowles et al.,
2016; Topaz, 2013; Topaz, Rao, Masterson Creber, & Bowles, 2013).
In the study reported here, we aimed to prospectively pilot-test
the PREVENT tool during hospital-to-homecare transitions. We tested
the PREVENT tool within a large homecare agency in New York City to
identify: (a) the tool's effect on the timing of the first nursing visit (i.e.,
whether high priority patients were given priority and visited ahead of
others) and (b) the association between PREVENT's use and the rate of
30 and 60 day rehospitalizations.
5
|
METHODS
This quasi-experimental pilot study was conducted in two phases.
First, in the control phase, the PREVENT score was calculated by the
researchers but not shared with the in-hospital homecare intake team.
The rationale for this choice was to establish the baseline comparison
group for the experimental phase. The in-hospital homecare intake
team included intake associates and RN homecare consultants—a
detailed description of the homecare intake team members’responsi-
bilities is provided in the next section and patient's homecare intake
workflow is described on Figure 1. Second, in the experimental phase,
the PREVENT score was calculated by the in-hospital homecare intake
team and shared with the homecare scheduling and assignment unit to
support patient prioritization. The study was compliant with the Health
Insurance Portability and Accountability Act (HIPPA) regulations on
data privacy. All patient data were stored on secure and password-
protected devices within the research organization. No identifiers
were disclosed to any outside entity. This study was approved by the
homecare organization's Institutional Review Board (IRB).
Before integrating the PREVENT tool into the workflow, we
discussed the potential implementation strategy and study design with
key organizational stakeholders involved with the intake and admis-
sion process. The stakeholders included the Senior Vice President of
Patient Care Services, the Director of Intake Services, a Regional Intake
Manager, an Intake Coordinator, an Interdisciplinary Care Team
Manager, and a Regional Director of Scheduling and Assignments.
These discussions helped us identify the key steps of the admission
process and study workflow.
5.1
|
Integrating the PREVENT decision support tool
into clinical workflow
The pilot test of the PREVENT tool was conducted during hospital-to-
homecare transitions in collaboration with an in-hospital homecare
intake team working within a large urban hospital in Brooklyn, New
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
FIGURE 1 Patient hospital-to-homecare transition in normal,
Q4
control, and experimental settings. Skilled need: a physician authorized
assessment that confirms the patient requires the skill of a registered nurse or physical therapist to perform and instruct in aspects of care
TOPAZ ET AL.
|
3
York during April and May 2016. Because homecare agencies differ
from one another in organizational structure, geographical location,
and other factors, we will explain the workforce structure of the
agency that participated in this study (see also Figure 1).
The regular course of transition from hospital to homecare started
with a referral from the attending physician based on the patient's
need for skilled care. Skilled care included rehabilitation services and
use of nurses to manage, observe, provide, and evaluate patient care.
Skilled nursing care was given or supervised by RNs. Referral from the
attending physician was received by an in-hospital homecare intake
team. Within the homecare intake team, an agency-employed
administrator, received the physician referrals and passed the
information to the homecare consultant, a RN, who also worked
within the hospital but was employed by the homecare agency. The RN
homecare consultant assessed the patient and affirmed the need for
skilled homecare. In the hospital, RN homecare consultants also had
access to patient charts, which ensured that the most current
information on the patient's health status was collected and
communicated. The patient's information was then passed on to the
homecare regional scheduling and assignment unit managers who
placed the patient into the scheduling system for the first home visit
with a RN (summarized in Figure 1). The time of the first nursing visit,
was considered the moment of admission to homecare.
The original plan for conducting the pilot study was to use the
PREVENT tool on 100 eligible patients in each phase of the study—control
and experimental as explained above. We excluded pediatric and post-
partum patients because the PREVENT tool was not designed or validated
for these patient groups. For the control phase, 100 consecutively
referred patients meeting the study criteria were included. We could not
calculate the PREVENT score for 10 patients because of missing or
incomplete clinical data. The control sample, therefore, included 90
patients. For the experimental phase, another 100 consecutively referred
patients fitting the criteria were included. For 14 patients in this group, the
PREVENT score could not be calculated because of missing or incomplete
clinical data, leaving a final sample of 86 patients.
To maintain blinding of the RN homecare consultants and visiting
nurses in the control phase of the study, a research assistant and the
primary investigator applied the PREVENT assessment tool and
calculated the PREVENTpriority scorefor a subset of patientspreviously
referred to homecare from the hospital. These scores were not shared
with the in-hospital homecare intake team. The data for completing the
PREVENT tool were collected from patients’medical records.
For the experimental phase, the investigators trained the four
hospital-based RN homecare consultants on how to complete the
PREVENT tool and calculate the priority score. The training included a
description of the tool development and validation processes. The
participating RN homecare consultants were also asked to calculate
patients’priorityusing the tool on a seriesof clinical scenariosreflecting a
diverse range of patients. Each of the clinical scenarios was then
discussed until agreement on the priority score was achieved by all the
participants. Any questions about the tool implementation or priority
calculation were discussed during the training session and also
communicated during PREVENT's implementation with the primary
investigator by telephone or electronic mail. The clinicians within the
scheduling and assignment unit responsible forthe timing of the patient's
first homecare visit by a nurse were also educated about the study.
Once all four RN homecare consultants were trained, they applied
the PREVENT tool and computed the score for 100 consecutive
patients referred to homecare during the experimental phase of the
study. Once the priority score was obtained by a RN homecare
consultant, the score was uploaded into the computerized referral
processing system. The priority recommendation (“high”or “medium/
low”priority for a first homecare nursing visit) was then automatically
sent to staff in the scheduling and assignment unit, who then
forwarded the priority recommendation along with the case assign-
ment to the nurse who made the first homecare visit.
5.2
|
Data collection
Patient-related information was collected from the electronic health
records of the homecare agency and included socio-demographic
characteristics (patient's age and gender), clinical characteristics
(primary and comorbid conditions captured as International Classifi-
cation of Diseases version 9 codes [ICD-9]), and episode of care
information (hospital discharge, the timing of the first homecare visit,
and, when applicable, rehospitalization). Additional PREVENT-specific
information also was collected from electronic health records including
the number of medications, the patient's need for toileting assistance,
and the presence of wounds.
The outcome measure of rehospitalization was defined as any
hospital admission that occurred between the patient's first homecare
visit until homecare discharge (up to 60 days). For the study purposes,
any rehospitalization was considered the end of the homecare episode.
Because rehospitalization data were extracted from the homecare
records, any and all rehospitalizations were noted, independent of the
hospital where the patient was readmitted. Time to rehospitalization
was calculated as the period between the patient's hospital discharge
date and the date of rehospitalization.
5.3
|
Data analysis
Appropriate statistical tests (bivariate association tests including t-tests
and chi-square tests) were conducted to compare the data from the
control and experimental phases. Weused Kaplan Meir survival analysis
to compare differences in time until rehospitalization (in days) for
patients in the control and experimental phases. Becausethis was a pilot
study with insufficient power to identify group differences in an
adjusted manner, all the comparisonswere unadjusted for other factors.
All statisticalanalyses were conducted in STATA v.11 (StataCorp, 2009).
6
|
RESULTS
6.1
|
General sample details
Characteristics of the patient sample for the control and
experimental study phases are reported in Table 1. The overall
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
4
|
TOPAZ ET AL.
study sample was two-thirds female with a mean age of 69.8 years
(SD = 17.3). On average, participants had 5.8 co-morbid conditions.
About 10% of the sample had a documented history of clinical
depression and 25% had wounds (skin impairment of any etiology).
Over half (59.7%) of the study sample required assistance when
toileting. Based on the PREVENT score, 25% of the study sample
was scored as medium/low priority for the first homecare nursing
visitandtherestwerehighpriority.Therewereafewstatistically
significant differences in patient characteristics between the
control and experimental study phase samples. For instance, the
experimental phase patients had slightly more co-morbid con-
ditions (6.1 vs. 5.4) and higher percentages of lipid metabolism
disorders and anemia. Conversely, control phase patients had
significantly more wounds (33.3% vs. 16.3%). Other differences
between the study samples in control and experimental phases
were not statistically significant.
6.2
|
Timing of the first nursing visit
During the control phase, there was no difference detected in time to
first nursing visit to homecare for both high and medium/low priority
patients. Both types of patients had a first nursing visit on average 2.2
days after hospital discharge. By contrast, in the experimental phase
high priority patients had a first nursing visit on average one-half a day
sooner than medium/low priority patients (1.8 vs. 2.6 days). These
differences were not statistically different although they have clinical
relevance.
6.3
|
Differences in rehospitalization during
homecare
Data on rehospitalizations were available only for the period in which
the patient received homecare services, so it is important to note the
homecare length of stay (LOS) for each phase. In the control phase the
average LOS was 24.7 days (SD = 17.9). In the experimental phase the
average LOS was 32.9 days (SD = 25.5). The LOS differences between
the two phases were not statistically significant. In the control phase,
21.1% (n= 19) of patients were rehospitalized during homecare versus
11.7% (n= 10) of patients in the experimental phase, a 9.4%
rehospitalization rate reduction for patients in the experimental phase.
Survival analysis, depicted in Figure 2, shows that risk of rehospitali-
zation was significantly higher (Log-Rank p= .025) in the control phase
versus the experimental phase.
7
|
DISCUSSION
In this study we examined the application of a clinical decision support
tool called PREVENT for homecare patient visit prioritization. Our
patient population was similar to the general homecare population
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
TABLE 1 Patient characteristics in the control and experimental study phase
Characteristics Total N= 176 Control phase n= 90 Experimental phase n=86
Female, n(%) 113 (64.2) 58 (64.4) 55 (62.7)
Age, M (SD) 69.8 (17.3) 69.3 (18.2) 70.4 (15.2)
Comorbid conditions
a
,n(%)
Hypertension 134 (76.1) 66 (76.7) 68 (75.6)
Cardiac morbidity 102 (58) 47 (54.7) 55 (61.1)
Diabetes mellitus 67 (38.1) 34 (39.5) 33 (36.7)
Urologic morbidity 62 (35.2) 30 (34.9) 32 (35.6)
Other nervous system 46 (26.1) 16 (18.6) 21 (23.3)
Lipid metabolism disorders 37 (21) 14 (16.3)* 22 (24.4)*
COPD 36 (20.5) 13 (15.1) 12 (13.3)
Anemia 20 (11.4) 6 (7)* 12 (13.3)*
Delirium, dementia, and other cognitive disorders 18 (10.2) 8 (9.3) 10 (11.1)
High-priority patients based on PREVENT, n(%) 132 (75) 65 (72.2) 67 (77.9)
Length of stay, days M (SD) 28.7 (22.2) 24.7 (17.9) 32.9 (25.5)
PREVENT items
Medications, M (SD) 9.5 (4.2) 9.7 (4.2) 9.5 (4.3)
Comorbid conditions, M (SD) 5.8 (1.7) 5.4 (.9)* 6.1 (2.4)*
Presence of depression
b
,n(%) 16 (9) 7 (7.8) 9 (10)
Presence of wound, n(%) 44 (25) 30 (33.3)* 14 (16.3)*
Limitation of toileting, n(%) 105 (59.7) 50 (55.5) 55 (61.1)
a
10 most common comorbid conditions are presented.
b
Depressive disorder, Not Elsewhere Classified COPD = chronic obstructive pulmonary disease.
*p< .01.
TOPAZ ET AL.
|
5
reported across the US with a few minor differences. For example,
patients in our sample had slightly more comorbid conditions, 5.8
versus 4 reported by Caffrey et al. (2011), and were slightly younger
(The National Association for Home Care and Hospice, 2010). With a
16.5% rate of 30-day rehospitalizations, our sample was similar to the
average rehospitalization rates in the US (The National Association for
Home Care and Hospice, 2010).
We also found a few differences between the control and
experimental study phases, the most significant being the number of
patients with wounds, which were 33.3% in the control versus 16.3%
in the experimental phase. We speculate that the difference might be
explained by a relatively lower rate of certain surgical procedures
conducted in the experimental phase. The experimental phase
coincided with the Jewish Passover holiday, during which Jewish
patients, who made up a significant proportion of the sample, are less
likely to schedule an elective surgery. In general, patients’reluctance to
undergo surgical procedures and be unable to leave their homes during
warmer months might account for the differences in wounds between
the control sample (data collected in the winter) and the experimental
sample (data collected in the spring). Otherwise, the differences
between the study samples were only minor.
Three-quarters of the patients in our sample were indicated as
high priority for the first homecare nursing visit based on the PREVENT
score. The PREVENT tool was developed based on data from a similar
healthcare system, however only about half of the sample in the
PREVENT tool development study were rated as high priority (Topaz,
2013). This might suggest differences in the severity of patient
populations between the studies. The large proportion of high-priority
patients in this study might also serve as an indication that additional
factors need to be taken into consideration in the patient priority
calculations by the PREVENT tool to improve its precision. On the
other hand, homecare organizations might need to allocate more
resources to respond effectively to large numbers of high priority
patients. To address these issues, further study with a larger sample is
needed to examine the sensitivity and specificity of the tool with
various populations.
In designing and using the PREVENT tool, we hope to help
homecare clinicians identify patients who are at increased risk of
poor outcomes, namely rehospitalization, so they may prioritize their
first homecare nursing visit during the critical transition period from
hospital to homecare. After application of PREVENT in the
experimental phase, we found that high priority patients were
admitted to homecare about a half day sooner than in the control
phase. In practice, this meant that high risk patients' problems were
potentially addressed earlier in the homecare episode. The
rehospitalization rates during homecare between the study phases
differed significantly, with 9.4% fewer rehospitalizations in the
experimental phase.
Several other studies of early interventions in homecare aimed
at reducing rehospitalizations also showed positive results. For
example, Murtaugh et al. (2016) showed that providing early and
intensive skilled nursing visits (also called frontloading) and early
physician follow-up in the first week of homecare services reduced
30-day rehospitalizations on average up to eight percentage points,
a 40% relative reduction. By contrast, several other studies have
found little or no benefits of early visits, potentially because patients'
priority was defined differently across agencies and implementation
of frontloading varied (O’Connor, Bowles et al., 2014; O’Connor,
Hanlon, & Bowles, 2014). We believe that better outcomes can be
achieved for homecare patients by combining standardized decision
support tools—like PREVENT- to aid decision making among
homecare nurses with other proven strategies such as ensuring
that patients see their primary care providers for follow-up visits
(Hernandez et al., 2010; Murtaugh et al., 2016).
Our study is not without limitations. This study was designed as a
feasibility pilot-study, and did not allow us to identify significant causal
relationships between the application of the PREVENT tool for
homecare patient prioritization, and patient rehospitalizations. The
quasi-experimental design did not support controlling for pre-post
differences or to do propensity scores or instrumental variable
analyses. Also, we applied the tool in one location, thus generalizability
to other locations and populations is limited.
We also did not conduct an in-depth examination of the inter-rater
reliability for nurses using the PREVENT tool. Further studies can also
explore the homecare admission processes in greater detail with
additional in-depth qualitative interviews with more managers, nurses,
and patients. In addition, data in the control phase were collected from
patients' charts by our study team, while data for the experimental
phase were collected by in-hospital RN homecare consultants, from
patients’charts and sometimes at the patients’bedside. This
distinction in data collection methods might have affected the
differences found between the control and experimental study
phases. Our study mainly focused on the implementation of the
PREVENT tool to streamline homecare admissions based on indica-
tions of severity/urgency of the patients, and rehospitalizations during
the homecare episode. Other factors at play during the homecare
episode like contact with and treatment by the primary care physician,
and content of the homecare treatment, were not taken into account in
this assessment.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
FIGURE 2 Survival plot of rehospitalizations in the control and
experimental study phases
COLOR
6
|
TOPAZ ET AL.
Importantly, we calculated rehospitalizations during the homecare
episode only, which might differ from general rehospitalization rates
outside of homecare. We were only able to follow patients during their
homecare episode, which ended after discharge from homecare. In
future studies, rehospitalization rates should be explored through
patient telephone follow-up, or from other data sources. Finally, we did
not account for differences in the admission times over weekends
given the low number of patients in our sample admitted on weekends.
8
|
CONCLUSION
We found promising results in this pilot study of patient prioritization
for the first homecare nursing visit. After applying and sharing the
PREVENT decision support tool with the homecare nurses, high
priority patients were seen sooner for the first homecare nursing visit
and rehospitalizations from homecare were significantly lower. Future
work is necessary to validate these pilot results using a larger sample in
a randomized controlled trial.
ACKNOWLEDGMENTS
This study was funded by the Eugenie and Joseph Doyle Research
Partnership Fund at the Center for Home Care Policy & Research,
Visiting Nurse Service of New York. We also thank the RINAH
reviewers and editorial team for their excellent guidance on these
manuscript revisions.
ORCID
Maxim Topaz http://orcid.org/0000-0002-2358-9837
REFERENCES
Barnason, S., Zimmerman, L., Nieveen, J., Schulz, P., & Young, L. (2012).
Patient recovery and transitions after hospitalization for acute
cardiac events: An integrative review. The Journal of Cardiovascular
Nursing,27,175–191. http://doi.org/10.1097/JCN.0b013e31823
9f5f5.
Bobay, K. L., Jerofke, T. A., Weiss, M. E., & Yakusheva, O. (2010). Age-
related differences in perception of quality of discharge teaching and
readiness for hospital discharge. Geriatric Nursing,31, 178–187. http://
doi.org/10.1016/j.gerinurse.2010.03.005.
Bowles, K. H., Ratcliffe, S., Potashnik, S., Topaz, M., Holmes, J., Shih, N.-W.,
& Naylor, M. D. (2016). Using electronic case summaries to elicit multi-
disciplinary expert knowledge about referrals to post-acute care.
Applied Clinical Informatics,7, 368–379. http://doi.org/10.4338/ACI-
2015-11-RA-0161.
Caffrey, C., Sengupta, M., Moss, A., Harris-Kojetin, L., & Valverde, R. (2011).
Home health care and discharged hospice care patients: United States,
2000 and 2007. (National health statistics reports; no 38). Hyattsville,
MD: National Center for Health Statistics. Retrieved from U.S.
Department of Health and Human Services website: https://www.
cdc.gov/nchs/data/nhsr/nhsr038.pdf.
Centers for Medicare & Medicaid Services. (2012). C.F.R. PART 484–home
health services. condition of participation: Comprehensive assessment of
patients. Washington, DC: U.S. Government Publishing Office. Available
online at: https://www.gpo.gov/fdsys/granule/CFR-2011-title42-
vol5/CFR-2011-title42-vol5-sec484-55/content-detail.html.
Centers for Medicare and Medicaid Services. (2016). Home Health Compare.
Available online at: https://www.medicare.gov/homehealthcompare/
search.html.
Coffey, A., & McCarthy, G. M. (2012). Older people's perception of their
readiness for discharge and postdischarge use of community support
and services. International Journal of Older People Nursing,8, 104–115.
http://doi.org/10.1111/j.1748-3743.2012.00316.x.
Foust, J. B., Vuckovic, N., & Henriquez, E. (2012). Hospital to Home Health
Care Transition: Patient, Caregiver, and Clinician Perspectives. Western
Journal of Nursing Research,34, 194–212. http://doi.org/10.1177/
0193945911400448.
Grafft, C. A., McDonald, F. S., Ruud, K. L., Liesinger, J. T., Johnson, M. G., &
Naessens, J. M. (2010). Effect of hospital follow-up appointment on
clinical event outcomes and mortality. Archives of Internal Medicine,170,
955–960. http://doi.org/10.1001/archinternmed.2010.105.
Hernandez, A. F., Greiner, M. A., Fonarow, G. C., Hammill, B. G.,
Heidenreich, P. A., Yancy, C. W., ... Curtis, L. H. (2010). Relationship
between early physician follow-up and 30-day rehospitalization among
Medicare beneficiaries hospitalized for heart failure. JAMA,303,
1716–1722. http://doi.org/10.1001/jama.2010.533.
Hosmer, D. W., Jr., Lemeshow, S., & Sturdivant, R. X. (2013). Applied Logistic
Regression. Hoboken, NJ: John Wiley & Sons, Inc., http://doi.org/10.10
02/9781118548387.
Jencks, S. F., Williams, M. V., & Coleman, E. A. (2009). Rehospitalizations
among patients in the Medicare Fee-for-Service program. New England
Journal of Medicine,360, 1418–1428. http://doi.org/10.1056/NEJMsa
0803563.
Ma, C., Shang, J., Miner, S., Lennox, L., & Squires, A. (2017). The Prevalence,
Reasons, and Risk Factors for Hospital Readmissions Among Home
Health Care Patients: A Systematic Review. Home Health Care
Management & Practice,25,59–94.
Markley, J., Sabharwal, K., Wang, Z., Bigbee, C., & Whitmire, L. (2012).
A community-wide quality improvement project on patient care
transitions reduces 30-day hospital rehospitalizations from home
health agencies. Home Healthcare Nurse,30,E1–E11. http://doi.org/
10.1097/NHH.0b013e318246d540.
McDonald, M. V., King, L. J., Moodie, M., & Feldman, P. H. (2008). Exploring
diabetic care deficiencies and adverse events in home healthcare.
Journal for Healthcare Quality: Official Publication of the National
Association for Healthcare Quality,30,5–12. https://doi.org/10.1111/
j.1945–1474.2008.tb01164.x.
McDonald, M. V., Peng, T. R., Sridharan, S., Foust, J. B., Kogan, P., Pezzin,
L. E., & Feldman, P. H. (2012). Automating the medication regimen
complexity index. Journal of the American Medical Informatics Associa-
tion,20, 499–505. http://doi.org/10.1136/amiajnl-2012-001272.
Medicare Payment Advisory Commission (2014). Report to the congress-
Medicare payment policy: Home health care services.
Q5
Available online
at: http://www.medpac.gov/docs/default-source/reports/mar14_ent
irereport.pdf.
Misky, G. J., Wald, H. L., & Coleman, E. A. (2010). Post-hospitalization
transitions: Examining the effects of timing of primary care provider
follow-up. Journal of Hospital Medicine,5, 392–397. http://doi.org/10.1
002/jhm.666.
Murtaugh, C. M., Deb, P., Zhu, C., Peng, T. R., Barrón, Y., Shah, S., ...Siu, A. L.
(2016). Reducing rehospitalizations among heart failure patients
discharged to home health care: Effectiveness of early and intensive
nursing services and early physician follow-up. Health Services Research,
52, 1445–1472. http://doi.org/10.1111/1475-6773.12537.
O’Connor, M. (2012). Hospitalization among Medicare-reimbursed skilled
home health recipients. Home Health Care Management & Practice,24,
27–37. http://doi.org/10.1177/1084822311419498.
O’Connor, M., Bowles, K. H., Feldman, P. H., St Pierre, M., Jarrín, O., Shah, S.,
& Murtaugh, C. M. (2014). Frontloading and intensity of skilled home
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
TOPAZ ET AL.
|
7
health visits: a state of the science. Home Health Care Services Quarterly,
33, 159–175. http://doi.org/10.1080/01621424.2014.931768.
O’Connor, M., Hanlon, A., & Bowles, K. H. (2014). Impact of frontloading of
skilled nursing visits on the incidence of 30-day hospital rehospitaliza-
tion. Geriatric Nursing,35, S37–S44. http://doi.org/10.1016/j.gerinur
se.2014.02.018.
Sharma, G., Kuo, Y.-F., Freeman, J. L., Zhang, D. D., & Goodwin, J. S. (2010).
Outpatient follow-up visit and 30-day emergency department visit and
rehospitalization in patients hospitalized for chronic obstructive
pulmonary disease. Archives of Internal Medicine,170, 1664–1670.
http://doi.org/10.1001/archinternmed.2010.345.
Sockolow, P., Ms, D., & Bass, E. J. (2016). Homecare Nurses’Decision
−Making During Admission Care Planning. Paper presented at Nursing
Informatics 2016: EHealth for All: Every Level Collaboration −From
Project to Realization, 225, 28–32. http://doi.org/10.3233/978-1-61
499-658-3-28.
The National Association for Home Care and Hospice. (2010). Basic
statistics about home care. Available online at: http://www.nahc.org/
wp-content/uploads/2017/10/10hc_stats.pdf.
Topaz, M. (2013). The Hitchhiker's Guide to nursing informatics theory:
using the Data-Knowledge-Information-Wisdom framework to guide
informatics research. Online Journal of Nursing Informatics,17(3), 1–6.
Available online at: http://ojni.org/issues/?p = 2852.
Topaz, M. (2014). Developing a tool to support decisions on patient
prioritization at admission to home health care (Doctoral Dissertation).
Available online at: https://repository.upenn.edu/cgi/viewcontent.cgi?
article = 3285&context = edissertations.
Topaz, M., Rao, A., Masterson Creber, R., & Bowles, K. H. (2013). Educating
clinicians on new elements incorporated into the electronic health
record: theories, evidence, and one educational project. Computers,
Informatics, Nursing: CIN,31, 375–379. http://doi.org/10.1097/NXN.
0b013e318295e5a5.
Weiss, M., Yakusheva, O., & Bobay, K. (2010). Nurse and patient
perceptions of discharge readiness in relation to postdischarge
utilization. Medical Care,48, 482–486. http://doi.org/10.1097/MLR.0
b013e3181d5feae.
Zisberg, A., Shadmi, E., Sinoff, G., Gur-Yaish, N., Srulovici, E., & Admi, H.
(2011). Low mobility during hospitalization and functional decline in
older adults. Journal of the American Geriatrics Society,59, 266–273.
http://doi.org/10.1111/j.1532–5415.2010.03276.x.
How to cite this article: Topaz M, Trifilio M, Maloney D,
Bar-Bachar O, Bowles KH. Improving patient prioritization
during hospital-homecare transition: A pilot study of a
clinical decision support tool. Res Nurs Health. 2018;1–8.
https://doi.org/10.1002/nur.21907
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
8
|
TOPAZ ET AL.
