nursing research framework through collaborations with a variety of researchers
from disciplines other than nursing science.
3 Examples of Nursing Research Using Real-World Data
3.1 Fall Prediction Based on Electronic Medical Record Data
The first example of real-world data analysis is the development of a prediction
model for falls based on EHR through collaboration with informaticians.
For hospitalized patients, falls are an important incident for nurses to address. It
causes increased discomfort and anxiety about the patient’s movements, as well as
secondary injuries such as fractures and trauma. While falls are preventable, the
human and material resources available for prevention are limited, and thus patients
at high risk of falls need to be screened. As an existing screening method, there is a
method in which a cut-off value is determined by scoring a patient by evaluating
their mobility ability, movement, fall history, in-use drugs, complications, mental
state, and so forth. These scales have been translated into various languages and are
used worldwide. However, it is difficult to easily and objectively evaluate fall risk in
a clinical setting. This is because nurses are already very busy checking various risk
assessment scales daily to evaluate patients in multiple ways and recording daily
nursing practices. There is also the problem of external validity that the performance
of scales originally developed in various countries and facilities may not be applicable to hospitals where they work. If it is possible to objectively grasp the patient’s
condition that can change daily and predict falls with high accuracy, fall occurrences
will be reduced.
To solve this problem, Yokota et al., a Japanese research group belonging to the
information department of a university hospital, examined whether or not information for fall prediction could be obtained from the hospital database. They attempted
to construct a fall prediction model that is highly likely to be collected at any facility
and is based on the patient’s condition and treatment on a daily basis (Yokota & Ohe,
2016). Yokota and his colleagues used a wide variety of medical information
included in the EMR. In addition, information on falls input independently of the
EMR was extracted from other databases and linked by patient ID. In this study, the
authors defined falls as “an event judged to be a fall for which a fall report was
created by medical staff at a clinical site.” This definition is the most convenient and
clinically objective. As an important predictor, they used the Intensity of Nursing
Care Needs. This is an index that substitutes for the severity of the patients that
nurses record daily in hospitals throughout Japan. Since nurses are to check the
Intensity of Nursing Care Needs daily, for example, regarding the presence or
absence of wound care, the degree of independence such as transfer, the presence
or absence of assistance, and so forth, there is a feature where information on the
patient’s condition is updated daily.
Real-World Data-Based Care Innovation: Lessons Learned from Nursing Science
85
from disciplines other than nursing science.
3 Examples of Nursing Research Using Real-World Data
3.1 Fall Prediction Based on Electronic Medical Record Data
The first example of real-world data analysis is the development of a prediction
model for falls based on EHR through collaboration with informaticians.
For hospitalized patients, falls are an important incident for nurses to address. It
causes increased discomfort and anxiety about the patient’s movements, as well as
secondary injuries such as fractures and trauma. While falls are preventable, the
human and material resources available for prevention are limited, and thus patients
at high risk of falls need to be screened. As an existing screening method, there is a
method in which a cut-off value is determined by scoring a patient by evaluating
their mobility ability, movement, fall history, in-use drugs, complications, mental
state, and so forth. These scales have been translated into various languages and are
used worldwide. However, it is difficult to easily and objectively evaluate fall risk in
a clinical setting. This is because nurses are already very busy checking various risk
assessment scales daily to evaluate patients in multiple ways and recording daily
nursing practices. There is also the problem of external validity that the performance
of scales originally developed in various countries and facilities may not be applicable to hospitals where they work. If it is possible to objectively grasp the patient’s
condition that can change daily and predict falls with high accuracy, fall occurrences
will be reduced.
To solve this problem, Yokota et al., a Japanese research group belonging to the
information department of a university hospital, examined whether or not information for fall prediction could be obtained from the hospital database. They attempted
to construct a fall prediction model that is highly likely to be collected at any facility
and is based on the patient’s condition and treatment on a daily basis (Yokota & Ohe,
2016). Yokota and his colleagues used a wide variety of medical information
included in the EMR. In addition, information on falls input independently of the
EMR was extracted from other databases and linked by patient ID. In this study, the
authors defined falls as “an event judged to be a fall for which a fall report was
created by medical staff at a clinical site.” This definition is the most convenient and
clinically objective. As an important predictor, they used the Intensity of Nursing
Care Needs. This is an index that substitutes for the severity of the patients that
nurses record daily in hospitals throughout Japan. Since nurses are to check the
Intensity of Nursing Care Needs daily, for example, regarding the presence or
absence of wound care, the degree of independence such as transfer, the presence
or absence of assistance, and so forth, there is a feature where information on the
patient’s condition is updated daily.
Real-World Data-Based Care Innovation: Lessons Learned from Nursing Science
85