MAIN BARRIERS FOR QUALITY DATA COLLECTION IN EHR

A Review

Rui Mendes

1,2

and Pedro Pereira Rodrigues

1,3

Faculty of Medicine of the University of Porto, Al. Prof. Hernâni Monteiro, 4200-319 Porto, Portugal

Faculty of Sciences of the University of Porto, Rua do Campo Alegre, s/n, 4169-007 Porto, Portugal

LIAAD - INESC Porto, L.A. & CINTESIS - Center for Research in Health

Technologies and Information Systems, University of Porto, Porto, Portugal

Keywords: Data quality, Data collection, Electronic health records.

Abstract: The volume of health data is rising and health information technologies which include electronic health

records are a promising solution, on data management and collection, to achieve greater quality outcomes.

However, they often cause errors instead of preventing them. To study the main barriers to high quality data

collection from electronic health records, a qualitative review study was conducted using 5 different

database engines having only considered data quality and documentation issues, opportunities and

challenges for proper data collection, electronic health records data and corresponding databases quality. It

were included 16 articles from which data availability, format, accuracy and data accessibility were the most

focused problems to address. Still, solutions are available: early recognition of those problems, well

structured and designed EHRs, standard coding use, periodic accuracy monitoring and feedback and broad

use of such systems for the most daily tasks possible, among others. Altogether they can improve EHR data

quality for everyday use.

1 INTRODUCTION

Technological progress, health services commitment

to their use, improvements in clinician’s skills and

willingness to use health information technologies

contribute to raise data quality to the point it can be

used for research investigation. The support of the

health information technology (HIT), with the use of

a well structured and designed electronic health

record (EHR), allows the possibility to process more

accurately, effectively and with more efficiency the

large amount of information being produced and

managed every day and translate it to a better quality

care. (de Lusignan and van Weel, 2006).As these

systems are set, we should need a better physician

and patient education and also a better clarification

of what kind of information is necessary and wanted

by both of them (Berner, 2005).

Quality improvement and error reduction are two

of the justifications for health care information

technologies. However, researchers evaluating the

problematic implementation of clinical information

systems often find situations where they’re

responsible for errors, instead of preventing them,

affecting high quality data collection (Stead, 2007).

On this paper we aim to review the main barriers

to high quality data collection from EHRs. Then we

expect to have a better understanding why and what

is possible to be made to achieve better outcomes.

2 SEARCH STRATEGY

AND STUDY SELECTION

A qualitative review study was conducted on the

found literature about the main barriers in high

quality data collection from EHRs.

The database research was held on Google,

Google Scholar, PubMed, Scopus, ISI Web of

Knowledge and ScienceDirect using the following

key-words: barriers, high quality data, data

collection, EHR, Electronic Health Records. At

PubMed the following MesH Terms were used -

“Medical Records Systems, Computerized”, “data

collection” – plus the key-words – “quality” and

“barriers”.

Some queries were applied in order to refine the

search. From a selection of 116 eligible articles (17

PubMed, 69 ScienceDirect, 2 ISI Web of

451

Mendes R. and Pereira Rodrigues P..

MAIN BARRIERS FOR QUALITY DATA COLLECTION IN EHR - A Review.

DOI: 10.5220/0003124104510454

In Proceedings of the International Conference on Health Informatics (HEALTHINF-2011), pages 451-454

ISBN: 978-989-8425-34-8

 2011 SCITEPRESS (Science and Technology Publications, Lda.)

Knowledge, 21 SCOPUS and 7 Google Scholar) a

title and abstract analysis was performed from which

a total of 35 articles were selected regarding data

quality collection and health information structures

required to data quality. Workshop analysis, forum

presentations, letters and papers which regarded

implementation issues and situation analysis reports

of a given institution, general perspectives on quality

care improvement following the use of electronic

health records and specific workflow analysis were

excluded. After full-text review of 27 articles

concerning data quality and documentation issues,

opportunities and challenges for proper data

collection, electronic health records data and

corresponding databases quality, a selection of 16

articles was made.

3 MAIN BARRIERS TO HIGH

QUALITY DATA COLLECTION

Some problems may arise along the path from

collecting raw data into useful information, called

information quality problems. Aiming to data quality

we should define it accordingly with the definition

given in total data quality management (TDQM) as

data fit for purpose by its consumers (Strong et al.,

1997a, de Lusignan et al., 2006; Cruz-Correia,

2009). On several occasions “information” normally

relates to both data and information; but “data”

usually refers to information in its early stages of

processing, and “information” to the product at a

later stage (Strong et al., 1997a).

The data flow process has several actors who

influence the quality information obtained from such

data at a later stage (Cruz-Correia, 2009). Having

the right information when and where it’s needed

comes with certain demands like the increased need

for its correct filtering, context-sensitive decision

support, legal and ethical guidelines regarding

obligations to obtain and use the information,

achieve real patient-physician expectations

regarding the use and usefulness of the information,

and enhancing data accuracy. Health care is an

information-based science. Many clinical practice

acts involve gathering, synthesizing, and acting on

information (Hersh, 2002). Since patient information

has traditionally been incomplete and fragmented,

lifelong EHR stands as a promising solution to

achieve complete and accessible information

(Berner, 2005).

3.1 Electronic Health Records

According with the International Organization for

Standardization (ISO) definition for EHR, it is a

repository of patient data in digital form, stored and

exchanged securely, and accessible by multiple

authorized users. It has retrospective, concurrent,

and retrospective information and its primary

purpose is to support continuing, efficient and

quality integrated health care (Häyrinen, 2008).

Information manufacturing process encompasses

three main roles: information producers, responsible

for generating and providing information;

information custodians, who provide and manage

computing resources for storage, maintenance, and

securing information and information consumers,

who access and utilize information for their tasks

(Strong et al., 1997a).

There are four major aspects known to

information quality and fifteen dimensions

underlying them (Strong et al., 1997b). These four

major characteristics related to high-quality data are:

intrinsic data quality, data quality context, data

quality representation and data quality accessibility.

A quality data problem is defined when any

difficulty is encountered along one or more quality

dimensions that turn data completely or largely unfit

for use (Strong et al., 1997b).

From the selected literature data availability (on

15 articles), data format (on 15 articles), data

accessibility (on 14 articles), data accuracy (on 12

articles) constitute the main barriers in contrast data

validation, revenue cycle management, auditing

(only on 6) and data cleansing (on 4) were less

focused.

3.1.1 Data Sources/Availability

There are difficulties associated in storing over time

large amounts (not necessarily better) of varied

information which often has conflicting or

ambiguous concepts across different computer

systems: lack of pieces of information, different

values or representations (formats or codes),

aggregated and non-aggregated impaired data

movement across the industry due to lack of

mapping and connecting different and inconsistent

sources of data, ineffective data collection

mechanisms for some required fields; errors at data

entry from users and no data entry validation

mechanisms at that point; lack on the use of

international terminologies and resulting poor

semantic interoperability.(Strong et al., 1997a;

Weiner, 2007; Häyrinen, 2008; Vaughan, 2009).

HEALTHINF 2011 - International Conference on Health Informatics

452

Common data dictionaries and data warehouses

are a current solution to distributed system problems

(Strong et al., 1997a). The alternative is constant

maintenance of data and systems to address

changing data requirements (Strong et al., 1997b).

To improve data aggregation should be considered a

standardized infrastructure and moving to a single

comprehensive controlled vocabulary for structured

data, making data transfer between different services

easier (Hersh, 2002; de Lusignan et al., 2006).

Dedicated technology and human resources are

necessary to monitor, catch, and correct errors at the

point of transfer (AHIMA, 2008). In the event of

systems failure, business continuity planning,

policies, and procedures for healthcare

documentation are fundamental assets for data and

documentation quality (AHIMA, 2008).

3.1.2 Data Format

There are four methods for data capture in EHRs:

entering data directly, including templates or screens

completed by the user; scanning handwritten

documents; transcribing text reports created by using

dictation or speech recognition; interfacing or

feeding data from other information systems, such as

laboratory systems, radiology systems, blood

pressure monitors, or electrocardiographs. Each one

of these methods has strengths and weaknesses that

may have an impact on data quality (AHIMA,

2008).

Direct data entry produces discrete, structured

data that can easily be analyzed and reported.

However, such data may be less accurate and

negatively impact the quality of documentation

(McDonald, 1997; AHIMA, 2008). On the other

hand much information is stored as unstructured,

narrative data. Such data are difficult to use reliably

in queries for several reasons, including among

others misspellings, synonyms, homonyms and

negation (Weiner, 2007). “Coded” data are needed

to better represent a clinical concept, since there are

many forms to represent it, giving the necessary

attention to coding systems dynamics – new codes

are added all the time without old ones being

removed (Strong et al., 1997a; de Lusignan and van

Weel, 2006; Häyrinen, 2008). At present, there isn’t

a single standard system for recording structured

data, a standard approach to coding and

classification (de Lusignan and van Weel, 2006).

3.1.3 Data Accuracy

An accurate electronic health record can eliminate

rework by capturing data once at the source and

presenting it for reuse as needed later on, but is

rarely achieved in practice (de Lusignan and van

Weel, 2006; Stead, 2007). Accuracy of system

documentation is normally calculated using two

measures: the proportion of documented

observations in the system that are correct (true) –

correctness; and the proportion of observations that

are documented – completeness (Berner, 2005;

Stead, 2007). Common causes for data inaccuracy

include placing a question in the wrong person’s

workflow; not allowing for clinically relevant

answers; reflecting what the physician ordered but

not what the patient really did; among other gaps in

information about care by providers who are not

using the system (Stead, 2007). Also establishing the

order of events and the time lapse between each one

is also problematic, especially when are used several

unsynchronized mechanisms to tell the time (Cruz-

Correia, 2009). Another problem comes when we

don’t know where and who entered such data.

Bayesian inference, the development of terminology

and minimal data set standards and also structured

data entry may improve data completeness (Strong

et al., 1997a; Berner, 2005; Weiner, 2007; Häyrinen,

2008).

3.1.4 Data Accessibility

Data accessibility (filtered by ethical issues like data

ownership, security, confidentiality and privacy) is

surely an obstacle to research investigation by third

parties, as this issue is still unclear, without access to

them analysts can’t do research and managers can’t

make decisions, like the unclear details about the

research methods employed by researchers, not

allowing studies replication (Strong et al., 1997a; de

Lusignan and van Weel, 2006; Kaplan and Harris-

Salamone, 2009). Structured notes allows easier

information retrieval; as when an information

system is used, and semantic tagging of information

is used (Häyrinen, 2008). Policies and procedures

development should also consider, data capture and

access control methods, determine when a record is

complete, auditing, evaluation and maintenance of

code sets, attend to which components refer to the

legal health record and privacy and security

regarding integrity issues as well (AHIMA, 2008).

These permissions are also barriers to accessibility

and affect the overall reputation and value of this

data (Strong et al., 1997b).

MAIN BARRIERS FOR QUALITY DATA COLLECTION IN EHR - A Review

453

4 MAIN FINDINGS

AND RECOMMENDATIONS

From the point of view of this work intrinsic data

quality, data quality context, data quality

representation and data quality accessibility were

identified as major data quality characteristics. Data

availability, data format, data accuracy and data

accessibility arise as major problems identified,

relating to high-quality data collection on EHRs.

There are solutions to solve such problems like early

recognition of development of those problems and

direct physician entry or physician entry control.

Also, structured encounter forms and well structured

and designed EHRs that include anticipatory

prompts and that allow data linkage and aggregation

to data consumers are part of the solutions available.

A broad use of such systems for the most daily tasks

possible without compromising the goal of

compliant documentation and standard coding use

are also to consider. Other relevant issues are

periodic accuracy monitoring and feedback, better

research methods explanation, evidence-based

guidelines, automated data capture from patient

information systems and others. If attended they can

help reducing data quality problems in order to

improve EHRs suitability for general everyday use.

REFERENCES

Ahima. (2008). Quality Data and Documentation for

EHRs in Physician Practice. Journal of AHIMA, 79,

43-48. Retrieved from AHIMA Body of Knowledge.

Berner, E. S., Moss, J. (2005). Informatics Challenges for

the Impending Patient Information Explosion. J Am

Med Inform Assoc, 12, 614-617.

doi:10.1197/jamia.M1873.

Cruz-Correia, R., Rodrigues, P. P., Freitas, A., Almeida, F.

C., Chen, R., Costa-Pereira, A. (2009). Data Quality

and Integration Issues in Electronic Health Records.

In: HALL, H. V. C. A. (ed.) Information Discovery on

Electronic Health Records. CRC Data Mining and

Knowledge Discovery Series.

de Lusignan, S., Hague, N., Van Vlymen, J. &

Kumarapeli, P. (2006). Routinely-collected general

practice data are complex, but with systematic

processing can be used for quality improvement and

research. Informatics in Primary Care, 14, 59-66.

Retrieved from: http://www.ingentaconnect.com/con

tent/rmp/ipc/2006/00000014/00000001/art00008.

de Lusignan, S. & Van Weel, C. (2006). The use of

routinely collected computer data for research in

primary care: opportunities and challenges. Fam.

Pract., 23, 253-263. doi:10.1093/fampra/cmi106.

Häyrinen, K., Saranto, K., Nykänen, P. (2008). Definition,

structure, content, use and impacts of electronic health

records: A review of the research literature. Int J Med

Inform., 77, 291-304. doi:10.1016/j.ijmedinf.

2007.09.001.

Hersh, W. R. (2002). Medical Informatics: Improving

Health Care Through Information. JAMA, 288, 1955-

1958. doi:10.1001/jama.288.16.1955.

Kaplan, B. & Harris-Salamone, K. D. (2009). Health IT

Success and Failure: Recommendations from

Literature and an AMIA Workshop. Journal of the

American Medical Informatics Association, 16, 291-

299. doi:10.1197/jamia.M2997.

Mcdonald, C. J. (1997). The Barriers to Electronic

Medical Record Systems and How to Overcome

Them. J Am Med Inform Assoc, 4, 213-221. Retrieved

from: http://www.ncbi.nlm.nih.gov/pmc/articles/

PMC61236/

Orfanidis, L., Bamidis, P. D. & Eaglestone, B. (2004).

Data Quality Issues in Electronic Health Records: An

Adaptation Framework for the Greek Health System.

Health Informatics Journal, 10, 23-36.

doi:10.1177/146045804040665.

Pawlson, L. G. (2007). Health Information Technology:

Does It Facilitate Or Hinder Rapid Learning? Health

Aff, 26, w178-180. doi:10.1377/hlthaff.26.2.w178.

Stead, W. W. (2007). Rethinking Electronic Health

Records to Better Achieve Quality and Safety Goals.

Annual Review of Medicine, 58, 35-47.

doi:10.1146/annurev.med.58.061705.144942.

Strong, D. M., Lee, Y. W. & Wang, R. Y. (1997a). 10

Potholes in the Road to Information Quality.

Computer, 30, 38-46. doi:10.1109/2.607057.

Strong, D. M., Lee, Y. W. & Wang, R. Y. (1997b). Data

quality in context. Commun. ACM, 40, 103-110.

doi:http://doi.acm.org/10.1145/253769.253804.

Vaughan, C. (2009). Three Barriers to Effectively Using

Information Stored in EHRs. HealthLeaders Media.

Retrieved from: http://articles.icmcc.org/tag/seconda

ry-data-use/

Weiner, M. G., Lyman, J. A., Murphy, S., Weineer, M.

(2007). Electronic health records: high-quality

electronic data for higher-quality clinical research.

British Computer Society. Retrieved from

IngentaConnect.

HEALTHINF 2011 - International Conference on Health Informatics

454