Download Leveraging Technology for Research

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts

Nonlinear dimensionality reduction wikipedia , lookup

Transcript
Leveraging Technology
for Research
Frances R Vlasses, PhD, RN, NEA-BC
Mary Malliaris, PhD,
Ida Androwich,PhD, RN, FAAN
Barbara Caspers, MSN, RN
Mary Dominiak, PHD, MBA, RN
Acknowledgment: AONE Seed Grant, 2006-07
Objectives
 Describe Technology solutions for
multi center research partnerships
 Explore, using existing datasets,
results to date on the relationships
between the nurse professional
practice environment, nurse manager
preparation and selected nursesensitive patient outcomes
A Natural Partnership
Loyola University Chicago
 School of Nursing
 Resources to support
research
 Mutual interest in
research questions
 HSM faculty
 Catholic university
Catholic Health Initiatives
 Striving for excellence in
Evidence-based practice
 Mutual interest in research
question
 Study variables are issues
of concern to organization
 Catholic healthcare system
Purpose
This study explored the
relationship between
Staff Nurse (SN)
education level and SN
perception of the
professional practice
environment (SPPPE).
Staff Perception of Professional
Practice Environment Scale (SPPPE)
 The SPPPE is a 38 item Likert scale (Iverson et
al, 2004) designed to measure the
characteristics of the professional nursing
practice environment
 The SPPPE measures 8 characteristics of PPE:
autonomy; Clinician-MD Relations; control over
practice; communication; teamwork; conflict
management; internal work motivation; and
cultural sensitivity
 Range of possible total score on SPPPE is 38152
CHI/Loyola University Chicago Research Collaborative
Staff Nurse Perceptions of their Work Environment
Williston ND u
u Tacoma, Lakewood,
Federal Way, Enumclaw WA
u
Carrington ND
u Baudette MN
Valley City ND u
u Breckenridge MN
Denville, Sussex,
Boonten Township, Dover NJ
u
Ontario OR u
Lincoln NE
u
u Durango CO
Joplin MO
u
Cincinnati OH (2 sites)
u
u
u Martin KY
Louisville KY
uu
Lexington KY (2 sites)
Berea KY
u Hixson TN
u
Sherwood AR
Morrilton AR uu
Chattanooga TN
u
Little Rock AR
Procedures
 General





Negotiated partnership between LUC and CHI
CHI senior management approval
LUC IRB approval
Permission to use SPPPE obtained
Step by step procedures developed to guide sites
 Technical
 Survey Monkey Web-links developed for each site
 CNO’s completed written demographic survey
 Sample Recruitment
 Participating hospitals recruited by researchers
via Webinar (3)
 IRB approved recruitment materials provided to
each site
 LUC researchers worked directly with site
coordinator
Methods
 Staff Nurses: Two Options:
 WEBSURVEY (two week data collection) or
PAPER AND PENCIL option
 SHORT SURVEY: questions about practice
environment, organization, demographics
 CHI REQUEST: 5 questions added on
Patient Centered care
 Chief Nurse Executive:
 BRIEF SURVEY VIA MAIL
Survey Monkey Example
1.
INTRODUCTION
Welcome to the Job Satisfaction, Work Climate, Unit Effectiveness and Staff Nurse
Retention Study. We appreciate your taking the time to complete this survey.
The purpose of this study is to examine the factors that influence staff nurse job
satisfaction and retention and staff nurses' perceptions of their work environment
including the climate and its effectiveness.
The survey should take no more than 30 minutes to complete. If you are unable to
complete the survey in one setting, you will be able to save your answers and
return to the site at a later time to complete the questionnaire.
The survey has 3 parts. The first, the Professional Practice Environment Scale, is a
38 item list of items that will ask your opinion about your practice environment.
The second consists of 5 questions related to your opinions regarding your ability
to practice Patient-Centered Care. The third will ask you questions about the
organization your work for, the unit you work on, and about yourself.
The study team believes that there is little, if any, risk to participating in the study.
All data will be kept confidential. You can not be identified from your survey. Your
questionnaire will be given a study code number to be used in the analysis of the
data.
By participating in the study and completing the two questionnaires, you are
indicating that you understand the purpose of the study and give your consent to
be in the study.
Benefits for Organizations



Involvement in nurse run research
supports Magnet standards of
excellence
Be among the first to forward the CHI
initiative of building academic/
service partnerships
Information from the study will be
shared at the end of the study
PROPOSED TIMELINE
For June, 2008 Data Collection**
 Decide to join us and Contact Aimee Steadman with
MBO name and # sites. DEADLINE 04/09/08
 Decide on electronic or pencil and paper format
and Contact Aimee Steadman. DEADLINE
04/09/08.
 Obtain Letter of Agreement from your organization
before 04/23/08
 Identify onsite study contact person (preferably
CNE) for June 2-16 data collection period
 Loyola team contacts each site
** There is opportunity for additional dates
Selected Data Mining Results
Dr. Malliaris
First Step
 Data Mining begins with a question
• First the question determines appropriate techniques. The
technique determines the data form.
Question
DM Technique
The data must be cleaned so
that all data the technique sees
is “good” data.
Data Form
Cleaning the Data
 Cleaning: only rows where all 38
questions had been answered were used
 Altering: for methods requiring flag
data, data responses grouped into two
categories: Agree-Strongly Agree &
Disagree-Strongly Disagree
Techniques
 We ran several data mining techniques
to help us find patterns in the data set.
 Association Analysis looks for things
that occur together
 Decision Trees, using a specified
target, displays the most important
variables in reaching the value of the
target
 Support Vector Machine also is used
to model the value of a target and
ranks the input variables in order of
importance
Association Analysis
• Groups together questions that have
similar response patterns
• Data must be in Flag format (that is,
only two responses)
• Results give statements of the form
If…Then… along with the likelihood
of occurrence
Association Analysis Example
Decision Tree
• This methodology requires that one
variable be designated as a
“target”; all other variables are
used to explain the response in the
target variable
• Again, we used data divided into
two categories of ASA and DSD
• The target was statement 9:
“manager who is a good manager
and leader”
How Good Is This Tree?
Support Vector Machine
 Support Vector Machine (SVM):
•Robust classification and regression
technique
•Maximizes the predictive accuracy of a
model without over-fitting the training data.
•Suited to analyzing data with very large
numbers of predictor fields.
•Requires a target variable
•Generates a list of variable importance
How Well Did SVM Do?
Comparing Three Techniques
 Association Analysis
•Top Rule: 1, 11, 12, 23
 Decision Tree
•Most Important Variables:
12, 1, 18
 SVM
•Most Important Variables:
12, 1, 23
Questions?