Clinician Adoption Patterns and Patient Outcome Results in Use of Evidence-Based Nursing Plans of Care

Abstract

Delivery of safe, effective and appropriate health care is an imperative facing health care organizations globally. While many initiatives have been launched in a number of countries to address this need from a medical perspective, a similar focus for generating evidence-based nursing knowledge has been missing [1]. This paper reports on a collaborative evidence-based practice (EBP) research initiative that adds nursing knowledge into computerized care protocols. Here, a brief overview of the study’s aims, purpose and methodology is presented as well as results of data analysis and lessons learned. The research team examined nurses’ adoption patterns of EBP recommendations with respect to activity tolerance using four-month patient data collected from a pilot hospital. Study findings indicate a need for more focus on the system design and implementation process with the next rollout phase to promote evidence-based nursing practice.

Introduction

As the focus increases on reinventing healthcare systems to be rid of the current ills of giving too little, too much and wrong care [2–5], it is unsettling that the nurse’s role remains invisible in most of the current process/method/structure system transformation initiatives [6–8], including Porter and Teisberg’s (2006) model of the “complete care cycle by medical condition” [9]. These models have in common the concepts of processes within venues of care, protocols based on “best practice” standards for assessing and caring for a patient with a medical condition along a continuum. In the ideal sustainable healthcare delivery systems of the future, optimum patient value will be dependent on efficient, effective interdisciplinary team work. Nursing’s contribution to identified measurable results and accountability for quality and costs needs to be clearly defined.

In 2002, Aurora Health Care (below denoted as Aurora), a large Wisconsin-based integrated health system, prepared to implement electronic health records (EHRs) across its delivery system. Aurora encountered an absence of evidence-based nursing content available to use within its EHR system. The research project reported here emerged from discussions on how best to address that void. The scope included the re-design of care protocols to cross the full care continuum using evidence-based content for the care team. The decision was made to adopt a rigorous research methodology to generate best practice content in lieu of the many barriers that impede healthcare professionals in obtaining and applying best evidence. The barriers to overcome included a lack of clinician competency for literature searches, study evaluation and implementation, unfavorable attitudes toward EBP, and less supportive external environments [10]. Accordingly, the Knowledge-Based Nursing Initiative was launched in 2004 by partnership among Aurora Health Care, Cerner Corporation, and University of Wisconsin-Milwaukee College of Nursing (ACW).

The ultimate goal of the ACW collaborative research initiative is to determine and improve nursing’s contribution to patient outcomes. The ACW collaborative is doing this through the infusion of best practice knowledge embedded in the workflow and decision support of an intelligent clinical information system and documenting nursing practice through structured data that enables outcomes analysis and continuous knowledge generation [11].

Methods

The methodology used throughout this initiative addresses the following four research aims:

1. to discover and generate the evidence-based nursing knowledge related to major clinical conditions;

2. to define a framework for representing that knowledge to clinicians to use in clinical practice;

3. to provide structured data using standard nursing terminologies for quality measurement, outcomes and research; and

4. to establish a formal process for outcome evaluation and continuous quality monitoring.

Knowledge Generation and Transformation

The methodological approach used for evidence gathering and synthesis as well as representing the synthesized knowledge within the EHR system with decision support tools is illustrated in Figure 1. The ACW team started the knowledge generation process by working with clinical nurse experts to “name” and rank the “phenomena of concern” that most commonly occurred in the care of patients with major health conditions. Once identified, the research team considered quality measures to be evaluated, which guided a systematic search and critical appraisal process to discover, organize, and synthesize the evidence content to align with the nursing process.

Figure1.

Knowledge Generation & Transformation Methods

The EBP content was divided into two types of nursing knowledge for representation – referential and actionable knowledge. Referential knowledge stands for a collection of descriptive synthesized information pertinent to a given phenomenon of concern (i.e., topic area), including a synopsis of the phenomenon, evidence-based practice recommendations with supporting rationale, strength of recommendations as well as flow charts. This referential knowledge can be accessed as a point of information by any team member through a reference knowledge database. The knowledge database further provides a full list of references analyzed with a level of evidence and web links to the abstract of original articles via PubMed.

Actionable knowledge transformed from the referential knowledge refers to a set of actions that are programmed into the clinical rules engine for the system in response to the given documented value or result. For example, if a nurse identifies that a patient needed “assistance” for Activity of Daily Living (ADL) prior to the admission, the rules engine will push this problem to the nurse by placing it on the problem list and presenting a plan of care for activity intolerance to the nurse to activate if appropriate. From end-users’ point of views, the actionable knowledge aligned with nurses’ decision-making work process is available at a screen level (such as structured assessment forms and pre-built plans of care for given phenomena of concern) with the reference text.

To deliver the greatest value to patients at the lowest costs, the evidence-based practice recommendations, thus, are being developed across the full continuum of care in venues that include the home and the community. To date, 22 nursing phenomena of concern have been developed into EBP recommendations for assessments, problem identification, interventions, and outcomes evaluation. Table 1 lists these phenomena and the venues completed for adult populations.

When transforming the referential knowledge to actionable knowledge, the ACW team employed the existing knowledge representation/sharing/reuse process adopted at the software company. In other words, expert system analysts closely worked together with the knowledge developer and nursing informatists to develop decision logics and flow charts in accordance with the synthesized knowledge. The EBP content was then embedded into structured forms with discrete data elements for patient assessments and pre-formulated care plans with recommended nursing interventions and evaluations. These front-end decision support tools designed for practicing nurses communicate with middleware that includes business logics and provides system interfaces to the back-end decision support system (such as clinical data repository, rules engine and knowledge database) to trigger reminders and task lists. It should be noted that, at the time of initiating this project, the research team did not focus on the development of computer-interpretable practice recommendations using existing formal guideline representation languages such as Asbru, EON, GLIF, GUIDE, PRODIGY, and PROforma [12]. The complexity of nursing knowledge and work process in nature impeded using the highly structured knowledge encoding and execution processes at this time.

Once the knowledge transformation is completed, both the referential and actionable knowledge are distributed to relevant clients to facilitate evidence-based practice for which each respective organization might customize the actionable knowledge according to their policies and resources.

Concept Representation

Since the ability to communicate, transfer and compare nursing data is essential to meeting the aims of this collaborative project, a dedicated terminology task group was built into project activities from inception. The terminology task group members include members from all the collaborating partners. This team has focused specially on concept representation and the application of terminology standards to the nursing data embedded under the four categories of assessment, diagnosis, interventions and outcomes.

The task group used international standards to guide the concept representation work, including (a) International Organization for Standardization ISO 18104:2003 Health Informatics – Integration of a reference terminology model for nursing [13], (b) International Classification for Nursing Practice (ICNP^®) [14], and (c) Systematized Nomenclature of Medicine Clinical Terms (SNOMED-CT) [15]. In 2003, ISO 18104:2003 was approved by ISO and serves as a higher-level reference model for nursing diagnoses and interventions. While the focus of the ISO 18104:2003 is on nursing, it has influenced development of other domain standards as well.

Whereas the terminologies have informed this project, at the same time, this project has enhanced the terminologies. The terminology task group has worked closely with these standards organizations and has directly informed and expanded the content coverage in SNOMED-CT and ICNP^®. This collaborative project provides an opportunity to use and test the ability of formal terminologies in the EHR to communicate and compare data for analysis. The use of these terminologies for knowledge representation in EHR applications that support evidence-based decision-making is ongoing work that represents a significant focus of this collaboration.

EBP Content Implementation

The implementation of the EHR system at Aurora occurred in two phases. That is, clinical documentation (including vitals, Intake & Output, forms-based clinical notes) and eMAR (electronic medication administration record) were implemented about a year before care planning and the evidence-based knowledge embedded. Aurora’s system-wide nurse practice council embraced using evidence-based content in practice and advocated embedding the EBP content into their clinical documentation system. The EBP content was introduced in the context of implementing general care planning with universal and specialized care plans constructed for Aurora. The first two EBP phenomena listed in Table 1 (Activity Tolerance and Medication Adherence) were bundled in 85 care plans, and thus were not viewed any differently from the others by the front line nurse.

Figure 2 shows a screen containing the content for Activity Tolerance in a plan of care as a nurse would encounter it. Nurses would use this screen in the process of assessing a patient’s progress, providing nursing interventions and evaluating nurse-sensitive outcomes.

Figure2.

Content for Activity Tolerance Integrated In a Plan of Care

In addition, the care planning module links documentation to nursing’s workflow and the “activities list” tool that allows nurses to see all the interventions from their assigned patients’ care plans on their shift. As interventions are completed and documented, the activity is removed from the list. Also, reminders were pushed to the nurse for any interventions, tasks or actions missed or not documented.

Creation of a Research Database

To continue to monitor the adoption patterns and examine patient outcomes with respect to the nursing phenomena, the research team initiated creating a nursing research database in the Aurora data warehouse (DW). It is important to know that the operational database involving sheer volumes of clinical data requires large scale database solutions for data processing, reporting and outcome measurement. Table 2 shows a process involved from data extraction to acquisition of a de-identified dataset to protect human subjects according to the Health Insurance Portability and Accountability Act (HIPPA) privacy rule. Approvals from Institutional Review Boards at the University of Wisconsin-Milwaukee and Aurora were obtained prior to this study. The de-identified dataset created through the computational data preprocessing was subjected to this descriptive study.

Table 2.

A Process of Dataset Preparation

(1)	Define the scope of data extraction with respect to the nursing phenomenon.
(2)	Write ETL (extract-transform-load) program.
(3)	Load the transformed data files into the Nursing Research database constructed in the DW.
(4)	Perform data exploration for data preparation.
(5)	Develop computational procedures for data modeling and preprocessing.
(6)	Apply a procedure to de-identify all the personal identifiable information from the database.
(7)	Create a de-identified dataset for process and outcome evaluation.

Study Findings

A pilot test was conducted at an acute care hospital with 72 beds, which was brought live in May 2006. Nurses use laptops on mobile carts for point of care documentation. A post-implementation descriptive data analysis was done using four months of data gathered from adult patients who were admitted from July through October 2006.

Sample Characteristics

The study population included a total of 1,141 medical, surgical and ICU patients (age ≥ 18). The majority of the subjects were female (59%) with mean age of 58 years (SD = 18.6, Median = 57). While medical and ICU patients (n=972) were admitted for emergent services (68%) with a mean length of stay of 4.6 days (SD =3.3, Mode=2), 87% of surgical patients (n=169) were admitted for elective surgery and stayed on average 3.1 days (SD =1.6, Mode=2).

Adoption Patterns

Since the purpose of the analysis was to see the extent that the embedded knowledge pertaining to Activity Tolerance was used by nurses, our approach to the adoption pattern analysis was to examine electronic documentation data by following a decisional work process implemented at the pilot site. The work process specific to Activity Tolerance involved three generic assessments, one focused assessment, seven nursing interventions and two nurse-sensitive outcome evaluations as shown in Table 3.

Table 3.

Nurses’ Documentation Pattern to the EBP Recommendations related to Activity Tolerance

Care Component	Percentage of Documentation
	Med.n=773	Surg.n=169	ICUn=199	Totaln=1,141
Generic Assessment on Adm ADL’s Prior to Admission ADL Activity Activity Intolerance Symptoms	86% 99% 65%	88% 98% 53%	89% 99% 83%	87% 99% 66%
Focused ADL Assessment Bathing, Continence, Dressing, Feeding, Toileting, Transferring	48%	30%	30%	43%
Interventions Universal Plan of Care	89%	93%	82%	88%
Outcome Measurement Activity level evaluation for discharged setting Functional status evaluation relative to baseline ADL Index score.	85% 85%	80% 80%	88% 80%	84% 84%

These results show that the nurses overall achieved a high level of compliance relatively quickly with the introduction of evidence-based care plans. The nurses were doing initial ADL (Activity of Daily Living) assessments at nearly 100 percent on admission. From the initial assessment data, it was assumed that 33% of the admitted patients (n=1,141) might have actual or potential activity intolerance on admission. Apparently, these patients stayed longer than patients without activity intolerance problem regardless of the patient type. Table 4 presents the results from chi-square and t-tests comparing the two groups in relation to age, LOS and readmission rate.

Table 4.

An Analysis of Descriptive Correlation

Group	Type of Patient
	Med.	Surg.	ICU
Patients with Potential/Actual Act. Intolerance Mean Age Mean LOS (days) Readmission Rate	n=279	n=20	n=82
	69.2±16.8* 6.0 ±3.6* 19.4%*	52.4±14.7 3.9 ±1.7** 0%	65.9±15.2* 6.5 ±5.4* 6.1%
Patients without Activity Intolerance Mean Age Mean LOS (days) Readmission Rate	n=494	n=149	n=117
	53.7±18 3.8±2.4 10.1%	48.9±15.9 3.0 ±1.5 0%	56.3±17.2 3.3 ±1.9 6.0%

^*p < .001,

^**p < .05

It should be noted, however, the documentation of follow-up assessments for patients with potential or actual activity intolerance dropped off to 43% (n=164). Further, the follow-up assessment was no longer continued after the first time of assessment in most of the patients. That is, only 17 out of 164 patients were assessed more than once for functional decline during the hospitalization. Accordingly, it was impractical to determine if deteriorations in functional status (which is calculated by summing the six ADL scores – Bathing, Continence, Dressing, Feeding, Toileting, and Transferring) contributed to the extended length of stay and readmission rate. Interestingly, while the assessment data were not completely documented, taking the next set of activating interventions and documenting against outcome measures presented a higher compliance over 80 percent. For example, it appeared that 85% of the medical patients were frequently evaluated for the ADL functional status indicator without being further assessed about the six ADL scores.

Discussion

Since most of data fields were newly embedded into the system through this collaborative initiative, the first focus in this study was looking at the adoption patterns using electronic data recorded after the implementation. The exploration of clinical data accumulated throughout their stay provided the richly descriptive insights about both the use of evidence-based content and documentation tool in the practice.

Overall, the documentation patterns indicate that the use of universal plan of care, including the activity related nursing interventions and outcome measurement, might promote unnecessary documentation for patients without the activity intolerance problem. This type of noisy data might contribute to the identified weak linkage between the nursing care components (i.e., assessment, problem identification, intervention, and outcome evaluation). In addition, it is thought that a lack of staff education about the EBP content for practicing nurses might hinder the clinicians’ adoption to the system and consequently failed to document continuous assessment data necessary for the patient group who presented the activity intolerance problem on admission.

Lessons learned from this study informed the three partners who decided to make changes before additional data are analyzed. These changes include some software redesign to adapt the nurses’ decisional workflows, redesigning the universal and specific problem framework, and increasing staff education about the phenomena.

Moving nursing into an evidenced-based practice depends upon the extension of best practice into care delivery. Thus, best practice knowledge needs to be reflected in the interventions and outcomes used to guide and inform care delivery by nurses and the care team. The study findings suggest a need for evaluating clinicians’ adoption patterns and patient outcomes in an iterative way in order to get to an optimum use of this powerful tool for delivering safe and effective care.

This study has limitations in that the de-identified dataset provided for this study included not the entire records documented throughout the hospitalization but limited data elements which were subjected to this analysis. More importantly, the dataset used for this study might not reflect the actual nursing practice performed at the point of care so that the derived information from the dataset should be carefully interpreted.

Conclusion

While this research initiative has many significant components and reports initial findings on nurses adoption of evidence-based decision support tools, they are presented here as an indication of ACW methodological approach. Further studies are needed to identify what difference the evidence-based content is making in the assessments, interventions and outcome measures selected by nurses, the nurse-sensitive outcomes achieved, and overall patient health and cost outcomes. It will then be possible to quantitatively describe nursing’s contribution to patient outcomes.

Table 1.

List of Selected Nursing Phenomena

Phenomenon of Concern	Venue
Activity tolerance	Acute Care
Medication adherence	Acute Care
Medication management	Home Care
Delirium, risk for	Acute Care
Delirium	Acute Care
Falls, risk for	Acute/Home
Falls, post-fall	Acute Care
Moderate sedation monitoring	Acute Care
Fluid overload, risk for	Acute Care
Fluid overload	Acute Care
Venous thromboembolism	Acute Care
Depression	Acute/Community
Discharge readiness	Acute Care
Knowledge deficit: Heart failure	Acute Care
Infection, central IV	Acute Care
Infection, peripheral IV	Acute Care
Urinary tract infection, risk for	Acute/Home
Dyspnea	Acute Care
Health promotion with hypertension	Community