Outcome Calculations Based on Nursing Documentation in the First Generation of Electronic Health Records in the Netherlands

Abstract

Objectives:
Previous studies regarding nursing documentation focused primarily on documentation quality, for instance, in terms of the accuracy of the documentation. The combination between accuracy measurements and the quality and frequencies of outcome variables such as the length of the hospital stay were only minimally addressed.

Method:
An audit of 300 randomly selected digital nursing records of patients (age of >70 years) admitted between 2013-2014 for hip surgery in two orthopaedic wards of a general Dutch hospital was conducted.

Results:
Nursing diagnoses: Impaired tissue perfusion (wound), Pressure ulcer, and Deficient fluid volume had significant influence on the length of the hospital stay.

Conclusion:
Nursing process documentation can be used for outcome calculations. Nevertheless, in the first generation of electronic health records, nursing diagnoses were not documented in a standardized manner (First generation 2010-2015; the first generation of electronic records implemented in clinical practice in the Netherlands).

Full text

Quelle: Paans, W., M. Müller-Staub, and W.P. Krijnen, Outcome
calculations based on nursing documentation in the first
generation of electronic health records in the Netherlands, in
NI16. 2016, IOS Press: Amsterdam.
Outcome calculations based on nursing
documentation in the first generation of
electronic health records in the Netherlands
Wolter PAANS a,1, Maria MüLLER-STAUB a,b Wim P. KRIJNEN c
a Research Group Nursing Diagnostics, Hanze University of Applied Sciences
b Nursing PBS
c Research Group Healthy Aging, health Care and Nursing Hanze University of
Applied Sciences
Abstract. Objectives. Previous studies regarding nursing documentation focused
primarily on documentation quality, for instance, in terms of the accuracy of the
documentation. The combination between accuracy measurements and the quality
and frequencies of outcome variables such as the length of the hospital stay were
only minimally addressed. Method. An audit of 300 randomly selected digital
nursing records of patients (age of >70 years) admitted between 2013-2014 for hip
surgery in two orthopaedic wards of a general Dutch hospital was conducted.
Results. Nursing diagnoses: Impaired tissue perfusion (wound), Pressure ulcer, and
Deficient fluid volume had significant influence on the length of the hospital stay.
Conclusion. Nursing process documentation can be used for outcome calculations.
Nevertheless, in the first generation of electronic health records, nursing diagnoses
were not documented in a standardized manner (First generation 2010-2015; the
first generation of electronic records implemented in clinical practice in the
Netherlands).
Keywords. nursing documentation, outcome calculation, nursing process,
orthopedic surgery, electronic health record.
1. Introduction
Studies addressing outcome calculations based on the Nursing Process Documentation
(NPD) in electronic health records (EHRs) are inadequate [1], and the influence of
nursing diagnoses on the length of stay (LOS) is unknown [2]. Reliable and valid
analyses on the LOS as a dependent outcome variable rely on accurately documented
nursing diagnoses, interventions, and background information for care planning and
evaluations [3].
The explanatory power of documented nursing diagnoses can be calculated based
on accurately stored nursing information in EHRs [4,5]. Accurate nursing (risk)
diagnoses can be employed for early detection in care plans to prevent patients’ health
1 Corresponding author, Hanze University of Applied Sciences, 9714 CE Groningen, The Netherlands;
E-mail: w.paans@pl.hanze.nl

Quelle: Paans, W., M. Müller-Staub, and W.P. Krijnen, Outcome
calculations based on nursing documentation in the first
generation of electronic health records in the Netherlands, in
NI16. 2016, IOS Press: Amsterdam.
complications. Early interventions to solve issues in nursing diagnoses may positively
effect hospital efficiency and, therefore, decrease hospital expenditures. Nursing care
must be documented in a standardized nursing language for valid outcome calculations
[5, 6]. Current developments of EHRs require the use of Standardized Nursing
Language (SNL). SNL describes the literature-based Nursing Process [7], which is
taught and implemented utilizing a standardized nursing language. It includes
assessment, nursing diagnoses, nursing interventions, and nursing outcomes that are
established in scientifically based nursing classifications [8].
Only on the basis of classification does the Nursing Process serve its purpose: an
application of scientific knowledge being appropriate to the clinical patient situation
that is defined and validated as concepts [9, 10].
2. Objectives
This study focused on the implementation of nursing documentation in EHRs in order
to evaluate the explanatory power of nursing diagnoses on the LOS in hip fracture
patients admitted for surgery in orthopedic hospital settings. The research question
was: “What is the predictive power of nursing diagnoses documented in electronic
health records on the depended variable Length of Hospital Stay (LOS)?”
3. Materials and Methods
A retrospective cross-sectional record audit was performed by using the D-Catch
instrument for the assessment of nursing documentation in EHRs. Two independent
data collectors performed the audit and came to a consensus on the scores based on the
D-Catch guidelines [11]. The predictive power of nursing diagnoses on the LOS was
subsequently calculated.
4. Sample and Population
An audit of 300 randomly selected digital nursing records of patients (age of >70 years)
admitted between 2013-2014 for hip surgery in two orthopaedic wards of a general
Dutch hospital was performed. Records were digitally archived and selected by digital
blind random sampling based on record number. The sample was selected from an
EHR software that is used in 50% (n= 45) of all Dutch hospitals (N= 90).
5. Data-analyses
With the D-Catch instrument, a total of 300 EHR’s of elderly patients (age > 70) with
hip fractures were examined. Measurements of nursing records according to the D-
Catch variables were carried out in three phases: admission phase, post-operative
phase, and the phase of discharge, i.e., the last day of a patient’s hospital stay. Inter-
rater reliability of the D-Catch instrument was calculated by using Cohen’s weighted
kappa. The Advanced Nursing Process documentation was subdivided into: a)
documentation in score lists such as delirium scores, Visual Analogue Scales for pain
measurement, SNAQ-scores for malnutrition, and scales for pressure ulcer
measurements; and b) Nursing Process Documentation in free text. In the EHR
software, no technical or content associations existed between score lists, nursing
diagnoses, nursing interventions, or nursing outcomes. Therefore, connecting and inter-
correlating nursing information required the researchers to use the D-Catch instrument
for nursing documentation auditing [11]. The aforementioned score lists and the
documentation in free text analyzed the prevalence of nursing diagnoses in the post-
operative phase by using consensus scores.

Quelle: Paans, W., M. Müller-Staub, and W.P. Krijnen, Outcome
calculations based on nursing documentation in the first
generation of electronic health records in the Netherlands, in
NI16. 2016, IOS Press: Amsterdam.
Multivariate logistic regression analysis explored independent explanatory factors
for the LOS. The independent variables included: medical diagnoses (co-morbidities),
medical treatment, and nursing diagnosis. The dependent variable was the length of the
hospital stay.
Less than 25 total rates of medical diagnoses, co-morbidities, medical treatments,
and nursing diagnoses were excluded from the analysis. There are 21 explanatory
variables and 262 cases included for further analysis. Thirty-eight cases were excluded
as the information in the record was not feasible to use for final analysis.
Bayesian Model Averaging was used for variable selection by averaging the best
models in the model class according to approximate posterior model probability. The
counts of the number of days hospitalized forms the response which is modeled by
Poisson regression including a dispersion parameter for increasing variance. After
controlling for other explanatory variables, the exponent of the estimated parameters
are interpreted as the rate ratio which is the expected number of days hospitalized
considering the diagnosis related to the number of days hospitalized without the
diagnosis.
6. Ethical considerations
The ethic committee of the hospital approved the research plan. To guarantee patients’
anonymity, nursing documentations were anonymized and coded.
7. Results
In most records, the admission and the discharge documentation were incomplete or
did not exist, and explanatory power calculations based on admission or discharge data
were not possible. The nature of nurses’ documentation in the current EHR’s is
narrative and unstructured with numerous redundancies. However, post-operative
nursing diagnosis documentation was determined to be feasible for use as a final
analysis.
Results are calculated from modeling the days hospitalized with Poisson
regression in terms of the estimated parameters, their standard errors (SE), t-value,
significance measured by the p-value, the rate ratio, and their 95% Confidence interval
(Table 1).
Analyses of electronically stored nursing diagnosis documentation revealed the
prevalence of post-operative nursing diagnoses that had a positive significant influence
on the LOS. The most prevalent nursing diagnoses were: Nausea, Acute pain, Deficient
fluid volume, Imbalanced nutrition less than body requirements, and Impaired skin
integrity (pressure ulcer). The number of nursing diagnoses documented in the EHR
also had a significant influence on the LOS; however, documented medical treatments
had no significant influence. Co-morbidities documented in medical diagnoses related
to the LOS were ascertained to significantly influence those patients experiencing
cardiac disease, stroke, and diabetes.
8. Discussion
New knowledge is required regarding the effects of SNL implementation, education,
and training on the quality of the outcome documentation; more important is to know if
SNL has an influence on actual care quality and patient outcomes.
It is unknown if SNL, assisted by computer tools, has an influence on actual care
quality and patient outcomes [10, 11]. To evaluate nursing outcomes, controlled
experimental studies are suggested. New studies on nursing-sensitive patient outcomes

Quelle: Paans, W., M. Müller-Staub, and W.P. Krijnen, Outcome
calculations based on nursing documentation in the first
generation of electronic health records in the Netherlands, in
NI16. 2016, IOS Press: Amsterdam.
should focus on teaching nurses how to use SNLs in practice and compare
documentation of the findings of EHRs with nurses’ perceptions and experiences.
Nursing diagnoses demonstrated having explanatory power on the LOS, and the
prevalence of nursing diagnoses was strongly related to the LOS. Yet, the assumption
is that the Nursing Process with valid assessments, evidence-based nursing diagnoses,
interventions, and nursing-sensitive patient outcomes based on SNL has not yet been
entirely implemented in the current generation of EHR’s. One of the difficulties for
nurses is how to make the transfer from their own reasoning process related to the
assessment of the patient into SNL, which defines patient care needs as nursing
diagnoses, nursing interventions, and nursing-sensitive patient outcomes, as the
documentation systems do not provide SNLs. Studies addressing effects of the use of
SNL by nurses in clinical practice are lacking, therefore, it is ambiguous whether using
SNL applied to Nursing Process documentation in the EHR leads to improved patient
outcomes, therefore, it is an important topic for future research.
Table 1. Results from modeling days hospitalized by Poisson regression in terms of the estimated
parameters, their standard errors (SE), t-value, significance measured by p-value, the rate ratio, and their 95%
Confidence interval.
Estimate
SE
t-value
P-value
RR
Estimate
CLL
CLR
(Intercept)
1,2688
0,3657
3,47
6,00E-04
3,5567
1,7312
7,2595
Age
0,0103
0,0043
2,3788
0,0181
1,0104
1,0018
1,019
Impaired tissue perfusion
(wound)
0,3423
0,0768
4,46
0,0000
1,4082
1,2091
1,6338
Pressure ulcer
0,2607
0,0808
3,2261
0,0014
1,2979
1,1059
1,5183
Deficient fluid volume
0,3464
0,0899
3,8546
1,00E-04
1,414
1,1828
1,6826
Diabetes
0,214
0,0672
3,1848
0,0016
1,2386
1,0843
1,4111
Dementia
-0,1984
0,0782
-2,5378
0,0118
0,8201
0,7021
0,9539
References
[1] K. Saranto & U.M. Kinnunen, Evaluating nursing documentation – research designs and methods:
Systematic review, Journal of Advanced Nursing, 65(3), (2009), 464-476.
[2] F. Germini, E. Vellone, G. Venturini & R. Alvaro, Nursing outcomes: instruments for
visualizing the effectiveness of nursing care, Professioni infermieristiche, 63(4), (2010), 205-210.
[3] J.M. Brokel, K. Avant & M. Odenbreit, The value of nursing diangoses in electronic health records. In
T. H. Herdman (Ed.), NANDA International nursing diagnoses: Definitions and classification 2012-
2014 (pp. 99-113), Wiley-Blackwell, Oxford, 2012.
[4] M. Bruylands, W. Paans, H. Hediger & M. Muller-Staub, Effects on the quality of the nursing care
process through an educational program and the use of electronic nursing documentation, International
journal of nursing knowledge, 24(3), (2013)163-170, doi: 10.1111/j.2047-3095.2013.01248.x.
[5] J.M. Welton & E.J. Halloran, Nursing diagnoses, diagnosis-related group, and hospital outcomes,
Journal of Nursing Administration, 35(12), (2005)541-549.
[6] A. Freitas, T. Silva-Costa, F. Lopes, I. Garcia-Lema, A. Teixeira-Pinto, P. Brazdil & A. Costa-Pereira,
Factors influencing hospital high length of stay outliers, BMC health services research, 12, (2012)265,
doi: 10.1186/1472-6963-12-265.
[7] B.J. Ackley & G.B. Ladwig, Nursing diagnosis handbook: An evidence-based guide to planning care
(10 ed.), Mosby/Elsevier, St. Louis, 2014.
[8] M. Müller-Staub & W. Paans, Diagnosis-Related Groups and Electronic Nursing Documentation: Risks
and Chances, Comput Inform Nurs, (2010), doi: 1097/NCN.0b013e3181fcf814.
[9] G. Keenan, E. Yakel, K. Dunn Lopez, D. Tschannen & Y.B. Ford, Challenges to nurses' efforts of
retrieving, documenting, and communicating patient care information, Journal of the American Medical
Informatics Association, (2013)20(2)245-251 doi: 10.1136/amiajnl-2012-000894.

Quelle: Paans, W., M. Müller-Staub, and W.P. Krijnen, Outcome
calculations based on nursing documentation in the first
generation of electronic health records in the Netherlands, in
NI16. 2016, IOS Press: Amsterdam.
[10] C. Urquhart, R. Currell, M.J. Grant & N.R. Hardiker, Nursing record systems: effects on nursing
practice and healthcare outcomes, Cochrane Database Systematic Review, 1, (2009) CD002099.
[11] W. Paans, W. Sermeus, R.M. Nieweg & C.P.van der Schans, D-Catch instrument: development and
psychometric testing of a measurement instrument for nursing documentation in hospitals, Journal of
Advanced Nursing, 66(6), (2010)1388-1400, doi: 10.1111/j.1365-2648.2010.05302.x.