AACN Advanced Critical Care

Volume 28, Number 1, pp. 12-15

© 2017 AACN

Technology Linda Harrington, RN-BC, PhD, DNP,

Today CNS, CPHQ, CENP, CPHIMS

Department Editor

Closing the Science-Practice Gap With

Technology: From Evidence-Based Practice to
Practice-Based Evidence
Linda Harrington, RN-BC, PhD, DNP, CNS, CPHQ, CENP, CPHIMS

T echnology in the digital age is rapidly ushering in a new era of evidence

for clinical decision-making intended to overcome some of the challenges
associated with current evidence-based practice (EBP). This issue of Technol-
ogy Today highlights some of these issues as well as opportunities associated
with practice-based evidence (PBE). PBE involves the electronic collection,
aggregation, and analysis of clinical data for the purpose of augmenting or
automating decision-making and achieving optimal health outcomes.1 Unlike
EBP, PBE includes analyses of accessible data from all patients, all actual or
potential diagnoses and comorbid conditions, all variations in treatments and
interventions, and all variables and confounders captured in an electronic
database or warehouse such as the electronic health record (EHR). PBE affords
the real-world view of health, illness, and health care versus a controlled
experiment and the shortcomings that are associated with that.

Evidence-Based Practice
EBP is intended to support clinical decision-making in nursing by integrating
the best evidence from well-designed studies with nurses’ knowledge and exper-
tise plus patient-specific data, preferences, and values.2 The gold standard of
evidence in EBP is the randomized controlled trial, which is powered for the
outcome of interest and based on research that by design involves a limited
number of variables with deliberate elimination of confounding variables, patient
inclusion and exclusion criteria, and a clearly defined treatment protocol.
Findings, or evidence, from this research are then often applied in situations
that include variables beyond those involved in the study and to patients who
do not meet the inclusion and exclusion criteria and/or are not on the study-
defined treatment protocol, all of which influence outcomes.
Several authors have pointed out the limitations of clinical trials. These include
a lack of representativeness of trial participants in characteristics such as age,
race, gender, treatment, and comorbid conditions.3 Some patients receive long-
term therapies such as aspirin, statins, `-blockers, and angiotensin-converting
enzyme inhibitors when the research did not cover decades of treatment.
A greater value also is placed on statistical significance than clinical signifi-
cance in research.3 Too often, you see published research where a P less than
.05 is directly related to the sample size versus actual significance. What about
well-designed studies that find improvement with a certain treatment but not

Linda Harrington is an Independent Consultant, Health Informatics and Digital Strategy, and Professor,
Baylor College of Medicine, One Baylor Plaza, Houston, Texas 77030 ([Link]@[Link]).
The author declares no conflicts of interest.
DOI: [Link]

12
Downloaded from [Link] by AACN on March 13, 2017
VO L U M E 2 8 • N U M B E R 1 • S P R I N G 2 017 Technology Today

in enough patients to rise to the level of sta- are analyzed and fed back into the organiza-
tistical significance? Don’t we want to know tion for the purpose of improving clinical
more about those patients where the treatment outcomes. This feedback mechanism is a key
made a difference? feature of PBE. Individual patients, as well
Although meta-analyses and systematic as clinicians, benefit from what is learned
reviews may appear to be the answer in over- through the mass of electronic data being
coming shortcomings of research evidence, captured and analyzed 24/7.
the issue of publication bias remains, where The case of Vioxx (rofecoxib) illustrates
studies with negative findings are less likely the power of data within EHR databases.
to be published in some journals.4 As an exam- When data on prescribing and dispensing of
ple, in a recent meta-analysis of studies funded the medication were combined with addi-
by the National Institutes of Health on psy- tional data from laboratory tests, pathology
chological interventions in depression from and radiology studies, diagnoses, procedures,
1972 to 2008, the efficacy of these interven- and patients’ demographics, an increased
tions was found to be overestimated.5 Fifty- risk of acute myocardial infarction and sud-
five grants were awarded, but only 42 (76%) den cardiac death was discovered.9 The drug
resulted in published articles. Reasons for was removed from the market, and finan-
nonpublication cited by investigators included cial settlements were paid by the pharmaceu-
thinking that the findings were not sufficiently tical company.
interesting to be published, distraction by other The Vioxx case shows the value of EHR
obligations, and practical issues. data in retrospective analyses, but how do
On a much larger scale, researchers found we make it more real time, shorten the gap
that publication bias involving both overrep- between science and practice, and provide
resentation and underrepresentation of statis- clinicians and patients the information they
tically significant results in a large sample of need when they need it and where they need
meta-analyses from the Cochrane Database it? Technology is providing solutions.
of Systematic Reviews.6 In a sample of 1106 Work is underway to create automated
meta-analyses comparing the efficacy of real-time, practice-based models that allow
treatment with either placebo or no treat- nurses to make timely decisions on prevention,
ment, studies reporting statistically signifi- early intervention, and more precise care,
cant results in the treatment had on average thereby improving outcomes. An example is
a 27% higher probability of being included. the development of an automated model to
When focusing on safety, the researchers found predict patients at risk for 30-day hospital
that studies reporting no evidence of adverse readmission and 30-day mortality by using
effects had on average a 78% higher proba- EHR data.10 Using EHR data on medical diag-
bility of being included in the meta-analysis. noses, socioeconomic factors, and health care
Importantly, publication bias was larger in utilization, a predictive model was constructed
meta-analyses that included older studies. to identify inpatients at risk, thus allowing
individualized prevention efforts.
Practice-Based Evidence Data serve as the basis of PBE. As a result,
PBE involves capturing the evidence being practice-based evidence is currently limited
routinely generated through the structures and by issues with documentation and the avail-
processes of patient care. Using real-time cap- ability of analytics talent. Certainly the qual-
ture and analyses of practice level data yields ity of the data being input into the EHRs
a method whereby the findings can be used will affect the quality of the evidence created
in clinical decision-making.7 An important by those data. Forced-choice data elements,
benefit is that PBE brings accumulated racial lack of ability to input free text, and other
and ethnic disparities data into the decision- usability issues decrease the accuracy and
making process more so than what is present comprehensiveness of data and therefore the
in research. evidence the data produces.
The focus of PBE is on the needs of clini- The biggest challenges of PBE relate to
cians, patients, and families who are making database structures and interoperability, things
decisions on health and health care. It is anal- that nurses do not see but that will influence
ogous to creating a learning organization.8 the evidence being produced.11 For example,
Data on clinical work that is done every day many EHRs have underlying data structures

13
Downloaded from [Link] by AACN on March 13, 2017
Technology Today W W W . A A C N AC C O N L I N E . O RG

that are unintelligible with repetitive data Will EBP be replaced by PBE? Answers will
instances. Think about the number of places continue to be found where the data exist that
in the EHR where the same thing can be doc- are not only needed to make the clinical deci-
umented, both the exact words and synonyms. sion of interest but will increase the ease, rel-
Also think about other professionals simultane- evance, and timeliness with which information
ously documenting the same or similar thing in based on the data can be accessed. Although
additional synonyms. This makes a messy data- some of the limitations of EBP were pointed
base and creates huge challenges for analytics. out earlier, PBE also has limitations, espe-
Interoperability refers to the ability of the cially related to the technology capabilities
EHR to connect information systems, both mentioned earlier.
technically and semantically. Such connection
produces a larger pool of data, allowing a bet- Technology and Data Needs
ter picture of patients and their experiences. Many of the efforts today using technology
A lot of effort is being expended on this, but to promote the use of scientific evidence use
we have a long way to go. Overcoming the solutions that involve EBP guidelines, stan-
challenges related to PBE is a worthwhile dardized order sets, or alerts. These may or
endeavor in getting us to a true data-driven may not be relevant to the patient and situa-
health care system and a healthier nation. tion at hand. The focus of EBP in reducing
variation in patient care should also be ques-
Nursing Implications tioned, as patients are not all alike.
Two general areas must be considered as Although EBP focuses on reducing practice
we move forward into PBE. One involves variation and thereby reducing data variation,
evidence as it relates to clinical decision- PBE thrives on both. Using complex data
making. The other involves technology and analytics, PBE methods are intended to con-
data needed to fulfill PBE. tinually scan databases to find ideal solutions
for individual patients in real time. All of this
Evidence for Clinical Decision-Making requires a sound digital strategy and effective
When it comes to clinical decision-making, data management.
nurses look for answers where answers exist,
especially answers that are readily accessible Conclusion
and reliable. Equally important is nurses’ EBP and PBE are both about improving clin-
contribution to finding answers where answers ical decision-making and outcomes. Although
do not exist. This contribution occurs through older technology such as the internet readily
analyzing current research evidence or devel- brings research and other evidence to clinicians’
oping PBE solutions. desktops, tablets, or mobile phones, newer
Although both EBP and PBE support clini- technology will increasingly bring real-time,
cal decision-making, nurses’ clinical judgment patient- and context-specific evidence to the
and reasoning remain critical necessities. There point of care 24/7 to inform health care deci-
may be no or insufficient research or data and sions. Then what? What will follow PBE?
information specific to a patient’s unique We will talk about that in the next Technology
needs. And, of course, there are always those Today column.
times when technology is “down” or unavail-
able, requiring nurses to rely solely on their
judgment and the evidence-based standards REFERENCES
already in place. 1. Harrington L. Clinical intelligence. J Nurs Admin. 2011;
41(12):507-509.
One of the challenges for PBE is the exis- 2. Melnyk BM, Fineout-Overholt E. Evidence-Based Practice
tence and use of standardized treatments or in Nursing and Healthcare. 3rd ed. Philadelphia, PA:
EBPs within EHRs.1 This challenge is espe- Wolters Kluwer/Lippincott Williams & Wilkins; 2015.
3. Sheridan DJ, Julian DG. Achievements and limitations
cially apparent where compliance with these of evidence based medicine. J Am Coll Cardiol. 2016;
guidelines is measured and reported. This strat- 68:204-213.
egy can thwart innovation and discovery. It can 4. Reveiz L, Chapman E, Asial S, et al. Risk of bias of ran-
domized trials over time. J Clin Epidemiol. 2015;68:
cause premature closure of data being captured 1036-1045.
and analyzed, whereby a better intervention 5. Driessen E, Hollon SD, Bockting CLH, Cuijpers P, Turner
EH. Does publication bias inflate the apparent efficacy
and outcome may be missed because the EBP of psychological treatment for major depressive disor-
hardwired into the EHR prevents it. der? A systematic review and meta-analysis of US

14
Downloaded from [Link] by AACN on March 13, 2017
VO L U M E 2 8 • N U M B E R 1 • S P R I N G 2 017 Technology Today

National Institutes of Health-funded trials. PLoS One. .[Link]/books/NBK53494/pdf/Bookshelf_NBK53494.pdf.

2015;10(9):e0137864. [Link] Accessed November 7, 2016.
/articles/PMC4589340/. Accessed November 7, 2016. 9. Graham D, Campen D, Hui R, et al. Risk of acute
6. Kicinski M, Springate DA, Kontopantelis E. Publication myocardial infarction and sudden cardiac death in
bias in meta-analyses from the Cochrane Database of patients treated with cyclo-oxygenase 2 selective and
Systematic Reviews. Statist Med. 2015;34:2781-2795. non-selective non-steroidal anti-inflammatory drugs:
7. Kovacs RJ. Using practice-based evidence to reduce nested case-control study. Lancet. 2005;365:475-481.
disparities in care: a virtuous cycle turns. J Am Coll 10. Khan A, Hook ML, Simgh M, Vollbrecht M, Malsch A,
Cardiol. 2015;66(11):1234-1235. Malone ML. Using an automated model to identify older
8. Institute of Medicine (IOM). The Learning Healthcare patients at risk for 30-day hospital readmission and 30-day
System: Workshop Summary (IOM Roundtable on mortality. J Patient Cent Res Rev. 2016;2(4):203-204.
Evidence-Based Medicine). Washington, DC: The 11. Andrews J. Precision medicine. Healthc IT News. 2016;
National Academies Press; 2007. [Link] 13(8):4-6.

15
Downloaded from [Link] by AACN on March 13, 2017
Closing the Science-Practice Gap With Technology: From
Evidence-Based Practice to Practice-Based Evidence
Linda Harrington
AACN Adv Crit Care 2017;28 12-15 10.4037/aacnacc2017331
©2017 American Association of Critical-Care Nurses
Published online [Link]

Personal use only. For copyright permission information:

[Link]

Subscription Information
[Link]

Information for authors

[Link]

Submit a manuscript
[Link]

Email alerts
[Link]

AACN Advanced Critical Care is an official peer-reviewed journal of the American Association of Critical-Care
Nurses (AACN) published quarterly by AACN, 101 Columbia, Aliso Viejo, CA 92656. Telephone: (800) 899-1712,
(949) 362-2050, ext. 532. Fax: (949) 362-2049. Copyright ©2016 by AACN. All rights reserved

Downloaded from [Link] by AACN on March 13, 2017