Comprehensive
Discharge Planning By Advanced Practice Nurses
Review and Recommendations
By
Steve Miller
This quantitative study performed
by Naylor et al. (1999) attempts to examine benefits enjoyed by elders when they
are discharge and followed by Advanced Practice Nurses (APN). The APN monitored
their course post hospitalization. In a prior study, the authors noted improved
outcomes when discharge planning and home health is coordinated by an APN. This
article addresses at least two research problems.
Research
Problem
Naylor
et al. (1999) note that home health resources and the family of chronically ill
patients have been stretched to near maximum capacity. It is clear that when
someone can observe elder patients and assist with their care they are admitted
less often and require less acute care resources. An additional problem
statement inferred is that patients are going home sicker and earlier from
hospital. Increased risk of preventable hospital readmissions and premature
nursing home placement are stated problems by these authors. Opening statements
by Naylor et al. (1999) provide this reader with adequate understanding of
problems experienced when elders are discharged without adequate discharge
planning. These problems are communicated to the reader effectively.
Constant
readmission, premature nursing home placement, and stressors placed on family
members of severely ill elders are an issue for nursing. Nursing, as a
profession, can improve patient outcomes and impact admission/readmission rates
when given the opportunity. Advanced practice nurses must develop protocols and
treatment plans that will diminish unnecessary admissions. Nursing leaders,
educators, and staff must demonstrate best practice in all aspects of care and
provide research-based information to hospital administrators demonstrating
positive effects on the bottom line, in discharge planning and all other aspects
that nursing can affect.
Hypothesis
Statement
Naylor
et al. (1999) provide a hypothesis based on prior research. The hypothesis
states, “this intervention would improve patient health outcomes and reduce
service utilization and health care
costs compared with usual hospital and home care.” (p. 614) This hypothesis is
reasonable, however, provides a broad statement based on one month of APN
directed discharge care. The research design calls for one month of APN care
after hospital discharge, the data seems to support the hypothesis but long term
potential and costs for such a program are ignored. With such a high drop out of
study participants one would have to question what long term benefits this
program may have. This hypothesis is directional; independent variables are
general, and easy to distinguish. Dependent variables are stated clearly for the
study. This hypothesis is stated in a research form, a null hypothesis does not
exist.
Literature
Review
Literature
review by Naylor et al. (1999) was inadequate. The authors mention their own
study briefly, and two other studies are mentioned in relation to outcomes
measures. Detail of Naylor’s prior study was not given, therefore, Naylor et
al. (1999) ask this reader to accept their perception of prior findings and
study type without data to support their assumptions. While this article
references other research in the “context” portion of page one notes, “but
the benefits of more intensive follow-up of hospitalized elders at risk for poor
outcomes after discharge has not been studied.” Essentially Naylor et al.
(1999) require the reader to accept their concepts prima fascia. The literature
review leaves this reviewer questioning findings in this study and parts of the
study design. Prior studies by Naylor et al. (1999) are not described in any
way, nor are the other studies that are mentioned.
Naylor
et al. (1999) make decisions in this study based on a prior study that cannot be
validated. If a reader wanted to employ APN discharge, they would have to seek
other related articles. The literature review is inadequate without explanation
for the lack of information.
Theoretical
and Conceptual Framework
No
specific grand theory or concept is stated. A reader can imply an overriding
concept that APN discharge planning is superior to standard visiting nurses. The
study would be stronger if a concept were stated that guides the study design.
Internal validity would be stronger with a stated concept or nursing theory
guiding the research. Individual care theory is briefly mentioned but is not
tied specifically to the outcomes measured. This along with the literature
review is inadequate.
Research
Design
This
experimental design with between group comparison is adequate. The mainly
posttest only, longitudinal design was done well. (A few variables such as
depression and functional status were collected pre and posttest.) The
intervention was then randomized to subjects selected for the study by a blinded
research assistant (RA). A convenience sample of elders was used to randomize
for intervention. Given the situation no other randomization for the
intervention could have been done.
The
intervention was well suited for the study question, however, the methods used
in the intervention leave many questions for this reviewer. Naylor et al. (1999)
make numerous references to the “protocol” without giving specific
information. The author does give time frames for home visits and sates
unlimited visits can be made. The protocol is not defined clearly. By failing to
define the protocol many rival hypotheses become possible, including the
likelihood that the regular visiting nurses could have had the same outcomes if
given the same resource availability.
It
is never stated what the standard discharge planning routine is for the
hospitals VN. Often very competent staff have less than optimal outcomes when
resources are minimal. With that thought in mind, I would like to see a form of
cross over or blocked study. Given the same circumstances, resources, and
physician availability could the standard visiting nurse have outcomes equally
as well as the APN staff did? The standard discharge protocol/routine should
have been given.
Hours
per DRG would have been interesting, this may have separated out the patients
most likely to be readmitted, and by tracking hours per patient, and DRG
reviewers could see if this protocol was realistic for practice.
Too little is known about the standard of care for the visiting nurses,
their protocols, or their training. Without this information one could question
the external validity of the dependent variables measured.
The
dependent variables that were measured were adequate for evaluation of the
stated hypothesis. With the journal space afforded for research articles versus
the voluminous work that goes into a research project I find these comparisons
very well done. The research problems, and the hypothesis are accurately
reflected by statistical and actual comparisons.
Internal
controls are adequate for the situation in the real world, many variables will
occur due to the lack of homogeneity in this sample. When dealing with
chronically ill elders it is not possible to isolate single DRG categories in
most cases. Co-morbid conditions are measured, removing one very large variable
within the subjects studied. The multivariate analysis of the data also offsets
this problem minimally. A more homogeneous sample would decrease variability.
Having data explaining the screening done by blinded research assistants also
would increase this reviewer’s confidence in the sample. The inclusion
criteria are somewhat vague with many general criteria including a
self-assessment of poor health. Many elders feel in poor health when they may
have anxiety or depression. Inclusion criteria could have been more specific to
limit variability of the sample.
Overall
design was good, information regarding the tools used to measure outcomes is
needed. Naylor et al. (1999) state, “Tools with known validity and reliability
were used”. The specific information about each tool is lacking. A tool
developed by the author for satisfaction is utilized, yet no data is given on
reliability or validity for this tool. Data was obtained by phone interview and
medical record review, both of which have an inherent degree of error.
This
study is externally valid to a limited degree. Not enough information is
provided regarding standards used by visiting nurses in the control group. The
information about the protocol is not sufficient. Stating there is no limit on
phone calls during specified hours, home visits, or teaching with the APN group
makes this reviewer suspicious of external validity. APNs may have exceeded what
is physically reasonable for most clinicians, real world resources are limited.
In a real practice setting there are self imposed, organizational, and fiscal
limitations. It is not possible to blind the APN to the current hypothesis. By
knowing the hypothesis, it is reasonable to expect the APN to exceed what is
normal or customary, in attempting to meet research objectives. Given the
protocol criteria specified it would not be possible to carry out this level of
care for long periods of time, for the aforementioned reasons the outcomes hint
of artificial findings. Therefore external validity is applicable to the stated
population, however, further studies that mimic real practice settings need to
be performed for greater external acceptance of these comprehensive discharge
practices.
Sampling
The
target population is defined clearly. The sample is drawn from patients that
match the study design, hypothesis, and outcome statements. Inclusion criteria
are broad and sweeping but accurate based on stated Medicare’s top 10
admission DRGs. This convenience sample is a non-probability sample. If
multi-center studies could be completed and true random selection achieved it
would improve external validity. Current sampling techniques were adequate to
enroll enough subjects into the sample to maintain sufficient power analysis.
An
RA who was blinded to the study hypothesis recruited subjects. The RA assessed
the patient and informed consent then was obtained. Using a computer algorithm
the research manager made random assignments of the selected subjects. Blinding
the RA and assigning subjects to groups by computer algorithm eliminates
bias in sampling. Twenty-eight percent of those interviewed were enrolled
in the study. Naylor et al. (1999) state this is consistent with other similar
sampling situations. As stated by the author, another strength of the sample is
their similarity in health status, sociodemographics, depression ranking, and
DRG. Attrition rate was similar in each group. Attrition groups and those
completing the study were very similar in all categories. The research team did
a very good job in picking their sample. In this reviewer’s opinion, it is
very appropriate to select subjects that are identified by the Medicare DRG top
admissions category. This tactic significantly improves external validity.
Therefore it is reasonable to generalize this study to other elders who have
similar DRGs.
Data
Collection
Data
collection was adequate within this study. Most of the data was collected by
telephone interview, or chart review. Both Methods have inherent problems.
Telephone interview allows for direct contact but also provides for
misinformation from study subjects. If subjects are aware of the study
hypothesis they may alter responses in an attempt to comply with expected
outcomes. When subjects died or
withdrew, the RA would abstract their information
from the medical record. Medical records are not always complete or accurate;
the author does state clearly that data collected prior to a subject’s death
or withdrawal was included in the study. Naylor et al. (1999) were careful to
use proven tools in the interview to assess depression, satisfaction and health
status. The subject’s physician and medical record verified readmissions and
co-morbid conditions. Blinding the RA who collected data appropriately controlled
for bias in data collection. There is no mention of bias due to anyone’s
presence when subjects were interviewed, no risk is assumed by the subject in
this study by providing information to an RA. Cost assessment data was based on
Medicare approved rates. This is not reflective or real cost. The intervention
group may have had greater cost savings than stated if real costs were used.
However, since both groups used the same Medicare cost criteria the between
groups ratio of cost should be correct. In
general the data collection was very well done with reasonable control for bias
and participant confidentiality. This portion of the study was done
appropriately given the circumstances.
Data
Quality
As
mentioned previously by this reviewer, the tools used for data collection in
this study are not well defined within the study. It would have been helpful to
show the tools if journal space was available. Specific validity and reliability
testing would have made a stronger case for these researchers when the actual
tool cannot be shown. Naylor et al. (1999) mention the tools are valid and
reliable. Future researchers will need specifics about the tools if this
research is to be replicated. Careful attention was given to data measurement
for readmissions, health status, depression, and follow up to the 24-week point.
Based on data collection and analysis it is appropriate for these authors to
assert the positive outcomes measured regarding the dependent variables.
Therefore, it is this reviewers opinion that the hypothesis was tested and found
to be true within the limitation I have mentioned.
Analysis
of Study Statistics
Naylor
et al. (1999) utilize many forms of advanced statistics. The use of the
statistics appears appropriate and extensive. Charts and explanations of
statistical data are acceptable. This reviewer would require additional
assistance from more advanced researchers to find fault with this authors use of
statistics. T-test and chi-square test were used appropriately. Difficulty in
understanding the extensive statistical tests is perhaps a weakness in the
research by Naylor. Average readers would need much help reviewing the
statistics as given. Charts provided by Naylor et al. (1999) help put the data
into perspective. Qualitative data from nurses and subjects would have been
helpful as well. Knowing the perceptions of the APN group regarding the protocol
could have added credibility to the positive findings of this study. Support for
the protocol by the APN would remove some skepticism regarding the ability to
apply this protocol to real world practice. Even without the qualitative data I
support the statistics within this study.
Evaluation
of Ethical Considerations
There
are many ethical considerations for this study. Subjects were required to be
English speaking and have a telephone. This would have excluded some of the
poorest potential subjects. It would be easy to assert that those subjects
excluded for lack of English speaking skills or phone availability would have
needed the most help. Other authors have shown that lower socioeconomic strata
have higher risk for illness and often-higher acuity upon admission to the
hospital. Perhaps for convenience the researches excluded a group most in need
of intensive discharge planning. Harm or risk to subjects was never an issue.
The non-intervention group received a visiting nurse, which was the current
standard of care. Coercion or deceit was not involved and subjects were allowed
to decline and withdraw easily as is indicated in the study data. Privacy was
maintained within reason, RA personnel calling for phone interview were the only
risk to privacy. It is noted that both hospital review boards approved this
study prior to its inception. The ethical considerations were reasonable and
appropriate based on the actions noted within this journal report.
Interpretive
Dimensions
The
researchers assertions clearly are supported by their data. Explanations for
findings are given in depth. The authors noted that 50% of the control group
patients not seen by VN immediately were readmitted quickly. With this in mind,
the authors show a five-fold decrease in readmissions when control subjects who
were seen by a VN immediately are compared to intervention group subjects. The
research problem and hypothesis are supported by study design and dependant
variable measurement. The authors suggest no functional status improvement
within the intervention group, and offer that improved coping with illness may
have been the result of APN intervention, not healthier subjects. There are no
rival hypotheses offered. Naylor et al. (1999) do suggest that the triggers for
referral to VN be studied as a potential causal link to high readmission rates.
No unwarranted outcomes are stated.
Presentation
Naylor
et al. (1999) do a very good job of presenting their study. The style of
writing, and presentation of the charts are very easy to read and understand.
The charts provided make the extensive statistical review easier to understand.
There is a potential bias in favor of Advanced Practice Nurses as indicated by
mention of the educational acumen possessed by an APN, yet the authors ignore
the unlimited resources the APN has. The bias is not overt, it does not affect
the outcomes. Mention of the APN should have been within the title. This article
was more to do with APN discharge planning than “Comprehensive Discharge
Planning”. In general this study was very well done.
Conclusion
and Recommendations For Change
This longitudinal, experimental, post-test only design was good, however, this design leaves some rival hypotheses not mentioned by the author. An example, Comprehensive discharge planning by Visiting Nurses decreases readmissions, reduces overall cost, and improves long-term health status of elders. It is my belief that given the same resources and time the Visiting Nurses would have near the same outcomes measured in this study. To eliminate this primary rival hypothesis I would design a Blocking design, two groups of protocol driven intervention with equal resources. One group would be APN, the other standard VN, a third control group of VN with routine protocol. This of course would be more expensive and difficult to get enough subjects for appropriate power analysis. I would also like to have seen hours per patient during the study with between group comparisons and statistics, this may have eliminated a need for blocking. Additional internal control for homogeneity would have helped the internal validity; a form of this was done when the authors compared readmissions with and without CHF patients included. Standard VN protocol would have been useful and should have been included. One could infer that the study hospitals have inadequate referral patterns/criteria (author noted also) or that the VN program was understaffed or inadequately trained. This study was very well done, the hypothesis drove the design and outcomes matched the data taken, statistical analysis applied to the data measured the hypothesis as stated. The authors chose their words carefully in making assertions with regard to dependent variables. Given the “real world” this study was exemplary.
References
Naylor,
M., Brooten, D., Campbell, R., Jacobsen, B., Mezey, M., Pauly, M., &
Schwartz, J. (1999). Comprehensive
Discharge Planning and Home Follow-up of Hospitalized Elders. JAMA, 281,
613-620.