An Ontological Map for Meaningful Use of Healthcare Information

Systems (MUHIS)

Arkalgud Ramaprasad

1, 2

, Thant Syn

and Mohanraj Thirumalai

School of Business Administration, University of Miami, 5250 University Drive, Coral Gables, FL, U.S.A.

College of Business Administration, University of Illinois at Chicago, 601 S Morgan Street, Chicago, IL, U.S.A.

Keywords: Ontology, Meaningful Use, Healthcare Information Systems, Electronic Health Records.

Abstract: An ontological map of meaningful use of healthcare information systems (MUHIS) is the visualization of its

requirements and practices using an ontology. We map (a) the Stages 1and 2 meaningful use requirements

set by the Centers for Medicaid & Medicare Services (CMS) for Electronic Health Records (EHR), and (b)

the current literature on meaningful use, to derive the ontological map of the requirements and practices

respectively. The map is fragmented and incomplete. The results will focus attention on the gaps (a) in the

requirements, (b) in practices, and (c) between requirements and practices, and highlight the bright, light,

blank, and blind spots in MUHIS. These gaps should be (a) bridged if they are important, (b) ignored if they

are unimportant, or (c) reconsidered if they have been overlooked. Feedback based on incremental

ontological maps over time will help to continuously improve MUHIS.

1 INTRODUCTION

Meaningful Use of Healthcare Information Systems

(MUHIS) is a work-in-progress at the national level

in the USA and other countries(Dermer and Morgan,

2010, Ke et al., 2012, Kim and Kim, 2012, Varroud-

Vial, 2011), at the local level in many states and

cities, and at the enterprise level in many hospitals,

physician practices, and other healthcare providers.

Its requirements and practices are evolving in

tandem, and along different paths depending on the

initial conditions, incentives, and the environment. It

is seen as an instrument for addressing the national

(USA, for example) concerns about the cost, quality,

and safety of healthcare. Consequently, there is a

constant pressure to continuously and rapidly

improve MUHIS. To catalyze the evolution, the

Centers for Medicaid & Medicare Services (CMS) in

the USA has set Stages 1 and 2 meaningful use

requirements for Electronic Health Records (EHR)

(Centers for Medicare & Medicaid Services). The

requirements specify the outcomes, associated

objectives, and corresponding measures. There are

incentives for meeting the objectives. Fulfilling the

requirements will be necessary but not sufficient for

harnessing the full potential of MUHIS; it has far

greater potential than envisioned in the present

requirements. The requirements and MUHIS have to

evolve quickly in tandem to meet the rapidly

increasing global demands on healthcare. It would

be a challenge to make the MUHIS ‘elephant’ dance.

MUHIS is a large, complex, and ill-structured

problem. It is a ‘wicked’ problem (Churchman,

1967). We have to manage its ‘wickedness’ through

feedback and learning to help it evolve rapidly. To

do so, we have to (a) abstract from the diverse, often

contradictory, and heterogeneous requirements and

practices of MUHIS, and (b) apply it to the

reformulation of requirements and practices

(Ramaprasad, 1979, Ramaprasad, 1983,

Ramaprasad, 1987). We need a clear framework

and method for abstraction and application

(Ramaprasad, 1987, Ramaprasad and Mitroff, 1984)

to avoid replaying the proverbial story of the five

blind men each of whom imagined an elephant as a

rock, an arrow, a fan, a rope, and a tree trunk after

touching its body, tusk, ear, tail, and leg,

respectively (Börner et al., 2003, Ramaprasad et al.,

2009). A wise man settles their argument about the

ontic nature of the elephant by piecing together the

picture for them. Fortuitously, the wise man in the

story could see and recognize the elephant; without

him the blind men’s argument would likely have

continued ad infinitum. Analogously we need wise

men and women who can see and recognize the ‘big

Ramaprasad A., Syn T. and Thirumalai M..

An Ontological Map for Meaningful Use of Healthcare Information Systems (MUHIS).

DOI: 10.5220/0004734500160026

In Proceedings of the International Conference on Health Informatics (HEALTHINF-2014), pages 16-26

ISBN: 978-989-758-010-9

 2014 SCITEPRESS (Science and Technology Publications, Lda.)

picture’ of MUHIS. The framework should guide the

abstraction, inform the application, and structure the

visualization of the MUHIS. It should thus help (a)

to limit the fragmentation of the requirements and

practices, (b) to make the system greater than the

sum of its parts, and (c) to evolve MUHIS

systemically and systematically. The framework

itself should be adaptable to the evolution of

requirements and practices through scaling,

extension, reduction, refinement, and magnification

of its components.

In the following, we will present an ontology

(Ramaprasad et al., 2009) for MUHIS and discuss a

method of mapping it using the framework. Thus,

we will present a map of the “knowledge structure”

(Zhang et al., 2012) of requirements and practices of

MUHIS as an ontological map. The ontological map

represents a “virtual knowledge

landscape” (Scharnhorst, 2001, p. 505) based on

textual empirical data about the requirements and

practices. It will help visually recognize the

coherence and lack of it in the cumulative domain

knowledge, and therefore help correct the lacuna

when appropriate (Hoeffner and Smiraglia, 2013,

Noar and Zimmerman, 2005). Thus, it will provide

“support for navigating the knowledge

landscape.” (Kazimierczak et al., 2012, p. 1) Further,

“[i]ncrementally computed information landscapes

are an effective means to visualize longitudinal

changes in large document repositories…" (Syed et

al., 2012, p. 352) such as the requirements and

practices of MUHIS. It will aid the continuous

improvement of MUHIS.

First, we will describe an ontology of MUHIS.

We will explain the conceptual foundations of the

framework and its bases in MUHIS requirements

and practice. We will also discuss the face, content

(Brennan et al., 2011), semantic (Kotis and Vouros,

2006), and systemic validity (Horn and Lee, 1989)

of the framework.

Second, we will describe the method for

mapping the requirements specifications and

practice literature onto the ontology and explain the

mapping process. We will discuss the reliability and

validity of the mapping. We will demonstrate the

visualization of the ontological map from the

mapped data based on the map.

Third, we will describe the gaps within

requirements, within practice, and between

requirements and practice using the ontological map.

We will discuss the importance of these gaps and

their implications for future requirements

specifications and practice.

Fourth, and last, we will describe how the

method can be used to develop incremental maps

(Syed et al., 2012) over time to generate feedback

and facilitate learning in the evolution of MUHIS.

We expect that continuous assessment and

improvement of MUHIS using the proposed method

will eventually lead to the realization of the guiding

vision.

2 ONTOLOGY OF MUHIS

Ontologies “… provide a shared and common

understanding of a domain that can be

communicated between people and heterogeneous

and widely spread application systems.” (Fensel,

2003, p.1) They “… make it possible to understand,

analyze, exchange or share knowledge of a specific

domain and therefore they are becoming popular in

various communities. However, ontologies can be

very complex and therefore visualizations can

support users to understand the ontology easier.

Moreover, graphical representations make

ontologies with their structure more manageable. For

an effective visualization, it is necessary to consider

the domain for which the ontology is developed and

its users with their needs and expectations.”

(Kriglstein and Wallner, 2013, p. 123)

Ontology is the study of being in contrast to

epistemology which is the study of knowing. Its

focus is on objects, their categories, and the

relationships between them. Ontologies represent the

conceptualization of a domain (Gruber, 2008); they

organize the terminologies and taxonomies of a

domain. An ontology is an “explicit specification of

a conceptualization.” (Gruber, 1995, p. 908) It is

used to systematize the description of a complex

system (Cimino, 2006). “Our acceptance of an

ontology is… similar in principle to our acceptance

of a scientific theory, say a system of physics; we

adopt, at least insofar as we are reasonable, the

simplest conceptual scheme into which the

disordered fragments of raw experience can be fitted

and arranged.” (Quine, 1961, p. 16)

There are potentially many ways of representing

a domain ranging from a natural-language narrative

to a formal mathematical formulation (when

possible). The ontology is a structured natural-

language representation, more formal than a

narrative but less formal than a mathematical

formulation. It is particularly suited for ‘wicked’

problems such as MUHIS. It is easy to understand

and apply the ontology.

The ontology for MUHIS is shown in Figure 1.

It encapsulates the logic of MUHIS. It has been

AnOntologicalMapforMeaningfulUseofHealthcareInformationSystems(MUHIS)

Figure 1: Ontology for Meaningful Use of Healthcare Information Systems (MUHIS).

formulated manually by the authors from the

meaningful use outcomes, objectives, and measures

(Centers for Medicare & Medicaid Services) and

their knowledge of the structure and functions of an

information systems. There is no computerized

method for extracting such an ontology (a) at this

level of granularity, (b) which is parsimonious (fits a

letter size page with legible font), and (c) has high

semantic validity (Kotis and Vouros, 2006) (each

combination is a natural English sentence as

explained below). During the formulation two of the

authors iterated between abstraction of the

framework from and its application(Ramaprasad and

Mitroff, 1984) to the requirements until the model

(a) was logically complete, and (b) covered all the

objectives, requirements, and criteria. It is similar to

the process described by Ramaprasad & Mitroff

(1984) and Ramaprasad (1987) for the formulation

of strategic problems.

The ontology has five columns representing the

five dimensions of MUHIS; two of the dimensions

together comprise the Health Information Systems.

Each dimension is defined by a one- or two-level

taxonomy. The dimensions are linked by

words/phrases interleaved between the respective

columns. The columns are ordered left to right such

that the concatenation of a word from each column

with the interleaved words/phrases results in a

meaningful natural English sentence. Four such

concatenated sentences are shown, with examples, at

the bottom of Figure 1. In the following we will

discuss the dimensions, the taxonomies, and the

concatenations in greater detail.

Management Structure Function Stakeholders Outcome

Analysis [of] Technology [for] Acquisition Recipients Efficiency

Specification Hardware Analysis Patients Quality

Design Software Interpretation Families Safety

Implementation Networks Application Population Disparities

Maintenance Processes Distribution Providers

Assessment Policies Physicians

Personnel Nurses

Pharmacists

Payers

Employers

Insurers

Regulators

Government

Examples:dataminingspecialists

HealthcareInformationSystems

[ofinformationby/to]

[tomeaningfullymanage]

1.Specificationoftechnologyforanalysisofinformationbyproviderstomeaningfullymanage

costofhealthcare.

Examples:electronichealthrecordssoftware,dataminingsoftware

2.Designofprocessesforacquisitionofinformationbypatientstomeaningfullymanagequalityof

healthcare.

FourIllustrativecomponentsofmeaningfuluseofHISfrom3360(6x4x5x7x4)level‐

1components:

Examples:accesstoonlinelabresults,formationofsocialnetworks

3.Implementationofpoliciesforapplicationofinformationbygovernmenttomeaningfully

managedisparitiesinhealthcare.

Examples:wellnesseducationpolicies,Medicaidreimbursementpolicies

[of/inhealthcare]

4.Implementation(deployment)ofpersonnelforinterpretationofinformationbyinsurersto

meaningfullymanagesafetyofhealthcare.

HEALTHINF2014-InternationalConferenceonHealthInformatics

2.1 Dimensions of the Ontology

The rightmost column is ‘Outcome’ and it lists the

four critical healthcare outcomes which need to be

meaningfully managed using HIS. They are

efficiency, quality, and safety of healthcare and

disparities in healthcare – a Core and Menu Set

outcome (Centers for Medicare & Medicaid

Services). There are many other Core and Menu Set

outcomes. We interpret them as means to the four

outcomes in the ontology. For example, consider the

Core outcome to ‘Engage patients and families in

their healthcare’. In the ontology patients and

families are stakeholders in achieving the desired

healthcare outcomes using the Health Information

System. Similarly, consider the Menu Set outcome

to ‘Improve population and public health’ – the

Population is a stakeholder receiving healthcare to

achieve the desired outcomes. We have been able to

relate all the present Stages 1 and 2 outcomes to the

four outcomes. In the future, additional outcomes or

subcategories of outcomes can be added, or some of

the outcomes deleted for application to a particular

context.

The second column from the right (Stakeholders)

is a taxonomy of stakeholders in HIS. They are the

recipients of healthcare (patients, families, and the

population as a whole), the providers of healthcare

(physicians, nurses, and pharmacists), payers for

healthcare, employers of recipients, insurers of

recipients, regulators of healthcare, and the

government. The categories of stakeholders are not

mutually exclusive – an entity may have multiple

roles. For example, a recipient may also be a payer,

and a self-insured employer may also be the insurer.

The categories may not also be exhaustive – they

may need to be extended or reduced. The present

taxonomy is a generic, parsimonious list of

stakeholders whose interests and roles in

meaningfully managing the healthcare outcomes

need to be considered.

The third and fourth columns (Structure,

Function) from the right are the common structural

and functional components of an information

system. They have been adapted to the CMS

terminology (Centers for Medicare & Medicaid

Services). The structural components of HIS are the

technology (hardware, software, and networks),

processes, policies, and personnel. The functional

components are acquisition, analysis, interpretation,

application, and distribution of information.

The leftmost column (Management) lists the

functions necessary to manage HIS to assure their

meaningful use. These are common functions in the

analysis, design, and assessment of any information

system. They are analysis, specification, design,

implementation, maintenance, and assessment; they

have been derived from the HealthIT terminology

(HealthIT.gov).

2.2 Illustrative Components

Each concatenation of words/phrases across the

framework is a potential component of MUHIS.

There are 3360 (6x4x5x7x4) level-1 and 7920

(6x6x5x11x4) level-2 components. We will focus

our discussion on the level-1 components and

subsume within them the details of the second level.

Four level-1 components are listed at the bottom of

Figure 1 with an example for each; they are

discussed below.

First, consider ‘Specification of technology for

analysis of information by providers to meaningfully

manage cost of healthcare.’ This could include

specification of cost-effective electronic health

records software to provide the type of clinical

decision support required by CMS for meaningful

use. It could also include data mining software to be

used by a large regional Health Maintenance

Organization (HMO) to determine the most

efficacious drugs for a commonly occurring chronic

condition.

Second, consider ‘Design of processes for

acquisition of information by patients to

meaningfully manage quality of healthcare.’ It could

include design of processes for online access of lab

results (required by CMS in Stages 1 and 2), or

processes to foster formation of social networks of

cancer patients to acquire information from each

other and form support groups.

Third, consider ‘Implementation of policies for

application of information by government to

meaningfully manage disparities in healthcare.’ It

could include policies to transmit health data to

government agencies, wellness education policies

for those living in the ‘food deserts’, and

reimbursement policies which help counter the

imbalance due to socio-economic status.

Fourth, and last, consider ‘Implementation

(deployment) of personnel for interpretation of

information by insurers to meaningfully manage

safety of healthcare.’ It could entail deployment of

data mining specialists to discover early warnings

about new potentially unsafe drugs.

Further, as shown in the examples, each

component may be instantiated in multiple ways.

The many instantiations constitute the MUHIS. We

note that some components may not be instantiated

AnOntologicalMapforMeaningfulUseofHealthcareInformationSystems(MUHIS)

at all in a given context. For example, without health

insurance the fourth illustrative component above

may be irrelevant. In general, the absence of

instantiation may reflect either an error of omission

(blind spot) or an irrelevant component (blank spot)

in that context.

2.3 Validity of the Ontology

In assessing the validity of the framework we note

that it is an ontology not the ontology for MUHIS;

we recognize that there can be other equally valid

frameworks. Each framework can be a lens to study

the domain; each lens can offer different insights

about the domain. Given that the MUHIS problem is

complex and ill-structured, ‘wicked’ (Churchman,

1967), a singular ontology is unlikely. We offer a

framework and its associated insights. It derives its

validity from its (a) logical construction, (b)

comprehensiveness, (c) interpretability, and (d)

completeness.

First, the logic of the MUHIS ontology’s

dimensions can be deconstructed as follows:

Meaningful Use of Healthcare Information

Systems = Meaningful Use + Healthcare

Information Systems

Meaningful Use = Management +

Stakeholders + Outcome

Healthcare Information Systems = Structure

+ Function

Thus, the dimensions comprehensively cover the

connotation of MUHIS. They can be easily

interpreted by a user.

Second, the categories of the taxonomy for each

dimension are logical and generally accepted in the

respective disciplines. Moreover, should a category

or subcategory be missing from a taxonomy, it can

be easily added. By the same token, a redundant

category or subcategory can be easily removed.

These corrections of potential errors of omission and

commission will not invalidate the rest of the

framework. Thus, the taxonomies of the dimensions

are comprehensive and interpretable.

Third, the ordering of the dimensions fits the

rules of English grammar – thus rendering the

concatenations in natural English and making them

meaningful and hence interpretable. Further, all the

components (concatenations) encapsulated in the

framework taken together provide a complete,

closed description of MUHIS.

Fourth, and last, the parsimonious representation

of the ontology provides a panoptic view of MUHIS

which can be analyzed with minimal cognitive

strain. A user can conveniently and meaningfully

explore its dimensions, elements, and components at

different levels of granularity.

Thus, we believe that the framework’s face

validity (Brennan et al., 2011), content validity

(Brennan et al., 2011), systemic validity (Horn and

Lee, 1989), and semantic validity (Kotis and

Vouros, 2006) are high. It parsimoniously

encapsulates the complexities of the system; it

makes the MUHIS ‘elephant’ known and visible and

hence can be used to map MUHIS systemically and

systematically. It is a simple, powerful tool to

synthesize and visualize the MUHIS knowledge

domain, to analyze the accumulation of knowledge

over time, and visualize its trajectory. It provides a

holistic approach to visualize the map and guide the

progress of a domain, for example, to answer the

question: How can we continuously improve

MUHIS? We explore these possibilities in the

following.

3 METHOD

3.1 Mapping MUHIS Requirements

We mapped all the Stages 1 and 2 requirements onto

the ontology through consensus mapping. The

requirements were obtained from the CMS website

(Centers for Medicare & Medicaid Services). Each

objective was mapped individually, considering it in

the context of the associated outcome and measures.

The total number of objectives mapped = 51. All the

objectives were first mapped by one author,

reviewed and modified by the other, and the

discrepancies between the two discussed and

resolved in the final mapping. The mapping does not

distinguish between the core and menu objectives,

and those for eligible professionals, eligible

hospitals, and CAHs (Critical Access Hospitals). We

provide two examples of mapping in the following.

Consider the Stage 1 core objective of

“Implement drug-drug and drug-allergy interaction

checks.” It is one of a set of objectives with the

stated outcome of “Improving quality, safety,

efficiency, and reducing health disparities.” We

mapped the objective for quality, safety, and

efficiency outcomes but not for disparities; we could

not see a direct link from the discussion of the

objective and its measures to managing disparities.

We mapped it to the ontology as: “Implementation

of technology/processes for application of

information by providers to meaningfully manage

efficiency/quality/safety.” We note that the objective

HEALTHINF2014-InternationalConferenceonHealthInformatics

corresponds to six components of the ontology, not

just one.

Consider the Stage 2 core objective of “Provide

patients the ability to view online, download and

transmit their health information within four

business days of the information being available to

the EP.” Although the stated outcome of the

objective is “Patient Electronic Access”, we inferred

the ultimate outcome to be primarily quality. It could

be efficiency and safety too, but we did not find

sufficient evidence to justify them. We mapped the

objective to the ontology as: “Implementation of

technology for distribution of information by/to

recipients/providers to meaningfully manage

quality.” Again, we note that the objective

corresponds to two components of the ontology.

Mapping the meaningful use objectives was

straightforward in most cases. It required little

interpretation except in the mapping the outcomes of

a few objectives as illustrated above. The mapping

was recorded on an Excel spreadsheet using one row

per objective and a column per element of the

ontology. All but 2 of the 51 objectives were

mapped to all the five dimensions of the framework

for a total of 65 full and 10 partial components.

3.2 Mapping MUHIS Practice

We mapped all the articles indexed in PubMed that

contain the term "Meaningful Use" in the

title/abstract and belong to the following MeSH

major topics: "Medical Informatics", "Medical

Records Systems, Computerized", "Electronic

Prescribing", and "Computer Communication

Networks". In addition, we also included articles

specifically designated to MeSH major topic

"Meaningful Use". The combined result was filtered

by date (2009 – March 1, 2013) as well as the

availability of abstract. We obtained a total of 200

articles. Of these, 43 were announcements,

editorials, etc. and 7 were non-US. They were

excluded from the study. The remaining 150 articles

were mapped by the authors onto the ontology based

on their titles and abstracts. Each article was mapped

by one author and validated by the other.

Differences in mapping between the two were

resolved through discussion. As with the

requirements an article could be coded (a) on all or

some of the dimensions, and (b) into a single or

multiple components of the framework. Of the 150

articles, 63 were coded on all the dimensions and 87

on a subset, for a total of 214 components and 1964

partial components. All the data were maintained

and mapped on spreadsheets (Google Docs and

Excel).

4 RESULTS

4.1 Ontological Map of MUHIS

Requirements

The ontological map of MUHIS requirements is

shown in Figure 2. The elements correspond to the

first level of the ontology. The number in

parenthesis adjacent to each element is the frequency

Figure 2: Ontological map of Stages 1 and 2 meaningful use requirements.

Management Structure Function Stakeholders Outcome

Analysis (0) Technology (49) Acquisition (21) Recipients (22) Efficiency (39)

Specification (0) Processes (44) Analysis (0) Providers (51) Quality (44)

Design (0) Policies (10) Interpretation (0) Payers (0) Safety (28)

Implementation (44) Personnel (0) Application (9) Employers (0) Disparities (9)

Maintenance (7) Distribution (28) Insurers (0)

Assessment (0) Regulators (0)

Government (7)

Healthcare Information System

[of]

[for]

[of information by/to]

[to meaningfully manage]

[of healthcare]

AnOntologicalMapforMeaningfulUseofHealthcareInformationSystems(MUHIS)

Figure 3: Partial histogram of Stages 1 and 2 meaningful use requirements.

Figure 4: Ontological map of meaningful use practice.

of its occurrence in the set of objectives. The bar

below the element is proportional to the frequency

using the total number of objectives (51) as the

denominator. The profile is very similar for Stages 1

and 2 requirements and hence they are not shown

separately. The total frequency for elements in a

column may exceed the total number of CMS

objectives due to one to many mapping of objectives

to components as illustrated and explained earlier.

The mosaic of the MUHIS requirements as a

whole is evident from the ontological map. It has

many bright spots (high frequency elements), light

spots (low frequency elements), and blank/blind

spots (no frequency elements). The no frequency

elements may be ‘blank’ by choice or ‘blind’ by

oversight – it cannot be resolved based on the data.

The mosaic may be summarized in a complex

sentence with higher frequency elements in bold and

decreasing frequency left to right as follows:

Implementation/ maintenance of

technology/ processes/ policies for

distribution/ acquisition/ application of

information by/to providers/recipients/

government to meaningfully manage

quality/efficiency/safety/disparities of

healthcare.

The partial histogram of Stages 1 and 2

meaningful use requirements shown in Figure 3

highlights the most common components of the

requirements using the structured construction of the

ontology – the bright spots. On the left is the

synthetic requirement based on the ontology, and on

the right the total frequency of its occurrence and a

proportional bar. As we have noted earlier, a CMS

requirement may be deconstructed into multiple

synthetic requirements. The full histogram (not

shown due to space constraint) portrays the bright,

light, and blank/blind spots at the component level,

in contrast to the element level visualization in the

ontological map.

Implementation of Technology for Distribution of information by/to Providers to meaningfully manage Quality of/in healthcare 19

Implementation of Processes for Distribution of information by/to Providers to meaningfully manage Quality of/in healthcare 17

Implementation of Technology for Distribution of information by/to Providers to meaningfully manage Efficiency of/in healthcare 16

Implementation of Processes for Distribution of information by/to Providers to meaningfully manage Efficiency of/in healthcare 16

Implementation of Technology for Acquisition of information by/to Providers to meaningfully manage Efficiency of/in healthcare 12

Implementation of Technology for Acquisition of information by/to Providers to meaningfully manage Quality of/in healthcare 12

Implementation of Technology for Distribution of information by/to Providers to meaningfully manage Safety of/in healthcare 11

Implementation of Processes for Acquisition of information by/to Providers to meaningfully manage Efficiency of/in healthcare 11

Implementation of Processes for Acquisition of information by/to Providers to meaningfully manage Quality of/in healthcare 11

Management Structure Function Stakeholders Outcome

Analysis (10) Technology (119) Acquisition (89) Recipients (15) Efficiency (32)

Specification (9) Processes (49) Analysis (6) Providers (124) Quality (50)

Design (11) Policies (15) Interpretation (5) Payers (2) Safety (18)

Implementation (87) Personnel (24) Application (41) Employers (1) Disparities (5)

Maintenance (1) Distribution (92) Insurers (0)

Assessment (66) Regulators (9)

Government (8)

[of]

[for]

[of information by/to]

[to meaningfully manage]

Healthcare Information Systems

[of/in healthcare]

HEALTHINF2014-InternationalConferenceonHealthInformatics

Figure 5: Partial histogram of meaningful use practice.

4.2 Ontological Map of MUHIS

Practice

The ontological map of MUHIS practice mirrors that

of the requirements, as one would expect, but with

the following significant exceptions: (a) in

Management there is a greater emphasis on

Assessment and virtually no emphasis on

Maintenance in practice; (b) in Structure, there is

less emphasis on Processes and slightly more

emphasis on Personnel in practice; and (c) among

Stake holders there is less emphasis on Recipients in

practice. The Function and the Outcome profiles of

the requirements and practice are similar.

The mosaic of the MUHIS practice as a whole is

evident from the ontological map. As with the map

of requirements it has many bright spots, light spots,

and blank/blind spots. The mosaic may be

summarized in a complex sentence with higher

frequency elements in bold and decreasing

frequency left to right as follows:

Implementation/ assessment/ design/

analysis/ specification/ maintenance of

technology/ processes/ personnel/ policies

for distribution/ acquisition/ application/

analysis/ interpretation of information by/to

providers/ recipients/ government/

regulators/ payers/ insurers to meaningfully

manage quality/efficiency/safety/disparities

of healthcare.

The partial histogram of meaningful use practice

shown in Figure 5 highlights the bright spots using

the synthetic components of the ontology. Its

construction is similar to Figure 3.

5 DISCUSSION

Words matter. The formulation of a problem can be

inclusive or restrictive, depending on the choice of

words and their connotations. We have formulated

meaningful use inclusively in the MUHIS ontology

(Figure 1). The Management dimension includes all

the major steps of a system development cycle; the

primary components of the Structure and Function

of a Health Information System are incorporated;

and so are all the key Stakeholders and Outcomes.

Meaningful use should be ‘meaningful’ for all the

stakeholders for all the key ‘uses’ (outcomes). The

inclusive formulation makes the MUHIS ‘elephant’

fully visible – doing so can diminish the costs of

fragmentation and drive the benefits of integration.

CMS has formulated meaningful use narrowly in

its Stages 1 and 2 requirements, as shown in the

ontological map in Figure 2. The narrow formulation

may be driven by its mission – their primary site for

MUHIS is ‘HealthIT.gov’ not ‘HealthIS.gov’,

emphasizing technology not the system. It may be

motivated by the strategy for implementation – to

start where there may be greatest leverage and to

proceed in stages. It may also be determined by their

decisions about their role. They may see motivating

recipients and providers as their role but not

motivating payers and employers. Similarly, they

may see motivating implementation as part of their

role but not analysis, specification, and design – the

latter could be the EMR vendors’ role. Last, the

similarity of Stages 1 and 2 maps suggest a

continuity of focus. We cannot adduce the reasons

for the map shown in Figure 2 and its continuity but

we can assert that the Stages 1 and 2 requirements

by themselves are unlikely to result in MUHIS in its

panoptic connotation expressed in the ontology. The

narrow formulation is likely to be suboptimal if not

dysfunctional.

Consider the Stage 1 objective: ‘Implement drug-

drug and drug-allergy interaction checks’. These

checks will directly affect the Quality and Safety

(Crosson et al., 2012, Rahmner et al., 2012, Spina et

al., 2011) Outcomes of healthcare (Classen et al.,

2011). Their effectiveness will depend upon the

providers’ response to the alerts issued based on the

Assessment of Technology for Acquisition of information by/to Providers to meaningfully manage Quality of/in healthcare 20

Assessment of Technology for Distribution of information by/to Providers to meaningfully manage Quality of/in healthcare 20

Implementation of Technology for Distribution of information by/to Providers to meaningfully manage Quality of/in healthcare 20

Implementation of Technology for Acquisition of information by/to Providers to meaningfully manage Quality of/in healthcare 18

Implementation of Technology for Distribution of information by/to Providers to meaningfully manage Efficiency of/in healthcare 18

Implementation of Technology for Acquisition of information by/to Providers to meaningfully manage Efficiency of/in healthcare 17

Implementation of Technology for Acquisition of information by/to Providers to meaningfully manage Safety of/in healthcare 14

Implementation of Technology for Distribution of information by/to Providers to meaningfully manage Safety of/in healthcare 14

Assessment of Technology for Acquisition of information by/to Providers to meaningfully manage Efficiency of/in healthcare 13

Assessment of Technology for Distribution of information by/to Providers to meaningfully manage Efficiency of/in healthcare 13

AnOntologicalMapforMeaningfulUseofHealthcareInformationSystems(MUHIS)

checks. Recent Assessment shows that more than

90% of the alerts are overridden due to alert fatigue

(Smithburger et al., 2011, Phansalkar et al., 2012b,

Crosson et al., 2012), information overload (Callen

et al., 2011), poor user interface Design (Seidling et

al., 2011, Gaikwad et al., 2007, Rahmner et al.,

2012), poor Specification of the critical interactions

(Gaikwad et al., 2007), and inadequate Analysis

(Phansalkar et al., 2012a, Takarabe et al., 2011) of

the interactions. It will be necessary to include most

of the blank elements in the map of Stages 1 and 2

(Figure 2) to improve the effectiveness of the

checks. First, it would be necessary to Assess

(Saverno et al., 2011, Warholak et al., 2011) the

current system to provide feedback (Smithburger et

al., 2011) for Analysis (Phansalkar et al., 2012a,

Takarabe et al., 2011), Specification, and Design of

the system. Second, the Assessment could be done

internally by a provider, locally, or by a conference

of all the Stakeholders (Phansalkar et al., 2012b,

Phansalkar et al., 2012a, Hines et al., 2012). Third,

any Assessment and feedback will entail extensive

Analysis (Phansalkar et al., 2012a, Takarabe et al.,

2011) and Interpretation (Dhabali et al., 2012) of

empirical data (Haueis et al., 2011). Thus, the

success of a large number of components

encapsulated in the ontology will be essential for

effectively implementing the ‘drug-drug and drug-

allergy interaction checks’. In absence of a

systematic systemic (Saverno et al., 2011)

perspective, the checks may be implemented but

they may be meaningless, especially if they are

overridden constantly (Yu et al., 2011).

6 CONCLUSIONS

The ontological maps and histograms provide clear

visualizations of the gaps within each and between

them. Some of these gaps definitely need to be

bridged, as in the case of decision support for drug-

drug and drug-allergy interactions. The policy

makers and practitioners have to assess the

importance of the other gaps and change

requirements and practices to bridge them. This

process of feedback and change has to be ongoing

for continuous improvement of MUHIS. Ontological

maps such as the ones presented in this paper can

provide the foundations for visualizing the domain,

monitoring the incremental changes, and making it

complete and integrated.

In summary, we present an ontological meta-

analysis and synthesis of MUHIS requirements and

practice (Ramaprasad and Syn, 2013). It highlights

the domain’s bright spots which are heavily

emphasized, the light spots which are lightly

emphasized, the blank spots which are not

emphasized, and the blind spots which have been

overlooked. It also highlights the biases and

asymmetries in MUHIS requirements and practice;

they can be realigned to make them stronger and

more effective.

As we have emphasized earlier our ontology is

one lens through which one can study MUHIS.

There can be other equally valid frameworks. Each

lens will likely yield a different map and thus

different insights into the bright, light, and

blank/blind spots. Each of these sets of insights will

be a product of observing the phenomenon

systematically through a systemic framework, of a

different way of making the ‘elephant’ visible.

Reconciling these differences, in addition to

changing the map of each will advance knowledge

of MUHIS and can set the research/practice agenda

for the domain.

The ontology is extensible and reducible, and

hence the method is adaptable to the developments

in MUHIS. Should a new Function or Stakeholder of

MUHIS emerge in the future, they can be added to

the framework. Or, should a new subcategory of

Providers becomes a key Stakeholder, the

framework can be extended to accommodate the

change. By the same token, if a category becomes

irrelevant, it could be eliminated from consideration.

The extensibility and reducibility will also help trace

the evolution of the constructs in and the logic of

MUHIS.

Last, but not the least, visualization is key to

making sense of and interpreting ‘big text data’ like

the emerging requirements and practice of MUHIS.

The ontology provides an easy and intuitively

understandable vehicle for visualization. Note, for

example, the ontological maps can be used to study

the evolution of MUHIS over time by creating maps

for different cross-sections of time. It can also be

used to study the map at different levels of

granularity using more refined/coarsened

taxonomies. These are works in progress. Feedback

based on incremental ontological maps will help to

continuously improve MUHIS. With the current

ontological map of MUHIS requirements and

practice it is unlikely that the full vision of

meaningful use will be realized – they have to

evolve a lot.

The evolution has to balance the emphasis on the

categories, dimensions, and components of the

ontology. It has to balance the bright, light, blank,

and blind spots. Following are three examples:

HEALTHINF2014-InternationalConferenceonHealthInformatics

 The emphasis on the Stakeholders has to be

balanced. All the stakeholders, individually and in

interaction with each other, collectively affect the

outcome.

 With the increasing role of ‘big data’ and data

mining in healthcare the low emphasis on Analysis

and Interpretation will likely have to increased

significantly

 Meaningful Use is itself a dynamic concept which

will evolve with time. MUHIS too has to be

equally dynamic. To do so the emphasis on

Analysis, Specification, Design, and Maintenance

(in Management) will have to be increased

considerably.

The ontological map can guide the evolution.

REFERENCES

Börner, K., Chen, C. & Boyack, K. W. 2003. Visualizing

knowledge domains. Annual Review of Information

Science and Technology, 37, 179-255.

Brennan, L., Voros, J. & Brady, E. 2011. Paradigms at

play and implications for validity in social marketing

research. Journal of Social Marketing, 1, 3-3.

Callen, J. L., Westbrook, J. I., Georgiou, A. & Li, J. 2011.

Failure to Follow-Up Test Results for Ambulatory

Patients: A Systematic Review. Journal of General

Internal Medicine, 27, 1334-1348.

Centers for Medicare & Medicaid Services. Meaningful

Use (Online). Available: https://www.cms.gov/

Regulations-and-Guidance/

Legislation/EHRIncentivePrograms/Meaningful_Use.

html.

Churchman, C. W. 1967. Wicked Problems. Management

Science, 14, B-141.

Cimino, J. J. 2006. In defense of the Desiderata. Journal

of Biomedical Informatics, 39, 299-306.

Classen, D. C., Phansalkar, S. & Bates, D. W. 2011.

Critical drug-drug interactions for use in electronic

health records systems with computerized physician

order entry: review of leading approaches. Journal of

Patient Safety, 7, 61-65.

Crosson, J. C., Schueth, A. J., Isaacson, N. & Bell, D. S.

2012. Early adopters of electronic prescribing struggle

to make meaningful use of formulary checks and

medication history documentation. The Journal of the

American Board of Family Medicine, 25, 24-32.

Dermer, M. & Morgan, M. 2010. Certification of primary

care electronic medical records: lessons learned from

Canada. J Healthc Inf Manag, 24, 49-55.

Dhabali, A. A. H., Awang, R. & Zyoud, S. H. 2012.

Clinically important drug–drug interactions in primary

care. Journal of clinical pharmacy and therapeutics.

Fensel, D. 2003. Ontologies: A Silver Bullet for

Knowledge Management and Electronic Commerce,

Springer.

Gaikwad, R., Sketris, I., Shepherd, M. & Duffy, J. 2007.

Evaluation of accuracy of drug interaction alerts

triggered by two electronic medical record systems in

primary healthcare. Health informatics journal, 13,

163-177.

Gruber, T. R. 1995. Toward Principles for the Design of

Ontologies Used for Knowledge Sharing.

International Journal Human-Computer Studies, 43,

907-928.

Gruber, T. R. 2008. Ontology. In: LIU, L. & OZSU, M. T.

(eds.) Encyclopedia of Database Systems. Springer-

Verlag.

Haueis, P., Greil, W., Huber, M., Grohmann, R., Kullak-

Ublick, G. A. & Russmann, S. 2011. Evaluation of

drug interactions in a large sample of psychiatric

inpatients: a data interface for mass analysis with

clinical decision support software. Clinical

Pharmacology & Therapeutics, 90, 588-596.

HealthIT.gov. Available: http://www.healthit.gov/

providers-professionals.

Hines, L. E., Malone, D. C. & Murphy, J. E. 2012.

Recommendations for Generating, Evaluating, and

Implementing DrugDrug Interaction Evidence.

Pharmacotherapy: The Journal of Human

Pharmacology and Drug Therapy, 32, 304-313.

Hoeffner, L. & Smiraglia, R. 2013. Visualizing Domain

Coherence: Social Informatics as a Case Study.

Advances In Classification Research Online, 23, 49-

51.

Horn, B. R. & Lee, I. H. Toward integrated

interdisciplinary information and communication

sciences: a general systems perspective. Proceedings

of the Hawaii International Conference on System

Sciences, 1989 Hawaii. IEEE, 244-255 vol. 4.

Kazimierczak, K. A., Skea, Z. C., Dixon-Woods, M.,

Entwistle, V. A., Feldman-Stewart, D., N'Dow, J. M.

O. & MacLennan, S. J. 2012. Provision of cancer

information as a “support for navigating the

knowledge landscape”: Findings from a critical

interpretive literature synthesis. European Journal of

Oncology Nursing, 1-10.

Ke, W. C., Hsieh, Y. C., Chen, Y. C., Lin, E. T. & Chiu,

H. W. 2012. Trend analysis and future development of

Taiwan electronic medical records. Stud Health

Technol Inform, 180, 1230-2.

Kim, H. & Kim, S. 2012. Legislation direction for

implementation of health information exchange in

Korea. Asia Pac J Public Health, 24, 880-6.

Kotis, K. & Vouros, G. 2006. Human-centered ontology

engineering: The HCOME methodology. Knowledge

and Information Systems, 10, 109-131.

Kriglstein, S. & Wallner, G. 2013. Human Centered

Design in Practice: A Case Study with the Ontology

Visualization Tool Knoocks. In: CSURKA, G.,

KRAUS, M., MESTETSKIY, L., RICHARD, P. &

BRAZ, J. (eds.) Computer Vision, Imaging and

Computer Graphics. Theory and Applications.

Springer Berlin Heidelberg.

Noar, S. M. & Zimmerman, R. S. 2005. Health Behavior

Theory and cumulative knowledge regarding health

behaviors: are we moving in the right direction?

AnOntologicalMapforMeaningfulUseofHealthcareInformationSystems(MUHIS)

Health Education Research, 20, 275-290.

Phansalkar, S., Desai, A. A., Bell, D., Yoshida, E., Doole,

J., Czochanski, M., Middleton, B. & Bates, D. W.

2012a. High-priority drug–drug interactions for use in

electronic health records. Journal of the American

Medical Informatics Association, 19, 735-743.

Phansalkar, S., van der Sijs, H., Tucker, A. D., Desai, A.

A., Bell, D. S., Teich, J. M., Middleton, B. & Bates, D.

W. 2012b. Drug–drug interactions that should be non-

interruptive in order to reduce alert fatigue in

electronic health records. Journal of the American

Medical Informatics Association.

Quine, W. V. O. 1961. From a Logical Point of View,

Boston, MA, USA, Harvard University Press.

Rahmner, P. B., Eiermann, B., Korkmaz, S., Gustafsson,

L. L., Gruvén, M., Maxwell, S., Eichle, H.-G. & Vég,

A. 2012. Physicians' reported needs of drug

information at point of care in Sweden. British Journal

of Clinical Pharmacology, 73, 115-125.

Ramaprasad, A. 1979. Role of Feedback in

Organizational-Change - Review and Redefinition.

Cybernetica, 22, 105-113.

Ramaprasad, A. 1983. On the Definition of Feedback.

Behavioral Science, 28, 4-13.

Ramaprasad, A. 1987. Cognitive Process as a Basis for

MIS and DSS Design. Management Science, 33, 139-

148.

Ramaprasad, A. & Mitroff, I. I. 1984. On Formulating

Strategic Problems. Academy of Management Review,

9, 597-605.

Ramaprasad, A. & Syn, T. 2013. Ontological Meta-

Analysis and Synthesis. Proceedings of the Nineteenth

Americas Conference on Information Systems,

Chicago, Illinois, August 15-17, 2013.

Ramaprasad, A., Valenta, A. L. & Brooks, I. 2009.

Clinical and Translational Science Informatics:

Translating Information to Transform Health Care. In:

AZEVEDO, L. & LONDRAL, A. R. (eds.)

Proceedings of HEALTHINF 2009 – Second

International Conference on Health Informatics.

Porto, Portugal: INSTICC Press.

Saverno, K. R., Hines, L. E., Warholak, T. L., Grizzle, A.

J., Babits, L., Clark, C., Taylor, A. M. & Malone, D.

C. 2011. Ability of pharmacy clinical decision-support

software to alert users about clinically important drug–

drug interactions. Journal of the American Medical

Informatics Association, 18, 32-37.

Scharnhorst, A. 2001. Constructing Knowledge

Landscapes Within the Framework of Geometrically

Oriented Evolutionary Theories. In: MATHIES, M.,

MALCHOW, H. & KRIZ, J. (eds.) Inegrative Systems

Approaches to Natural Social Dynamics.

http://www.virtualknowledgestudio.nl/staff/andrea-

scharnhorst/documents/constructing- knowledge-

landscapes.pdf: Springer.

Seidling, H. M., Phansalkar, S., Seger, D. L., Paterno, M.

D., Shaykevich, S., Haefeli, W. E. & Bates, D. W.

2011. Factors influencing alert acceptance: a novel

approach for predicting the success of clinical decision

support. Journal of the American Medical Informatics

Association, 18, 479-484.

Smithburger, P. L., Buckley, M. S., Bejian, S.,

Burenheide, K. & Kane-Gill, S. L. 2011. A critical

evaluation of clinical decision support for the

detection of drug-drug interactions. Expert Opinion on

Drug Safety, 10, 871-882.

Spina, J. R., Glassman, P. A., Simon, B., Lanto, A., Lee,

M., Cunningham, F. & Good, C. B. 2011. Potential

Safety Gaps in Order Entry and Automated Drug

Alerts: A Nationwide Survey of VA Physician Self-

Reported Practices With Computerized Order Entry.

Medical Care, 49, 904-910.

Syed, K., Kröll, M., Sabol, V., Scharl, A., Gindl, S.,

Granitzer, M. & Weichselbraun, A. 2012. Dynamic

Topography Information Landscapes–An Incremental

Approach to Visual Knowledge Discovery. Data

Warehousing and Knowledge Discovery, 352-363.

Takarabe, M., Shigemizu, D., Kotera, M., Goto, S. &

Kanehisa, M. 2011. Network-Based Analysis and

Characterization of Adverse Drug–Drug Interactions.

Journal of chemical information and modeling, 51,

2977-2985.

Varroud-Vial, M. 2011. Improving diabetes management

with electronic medical records. Diabetes Metab, 37

Suppl 4, S48-52.

Warholak, T. L., Hines, L. E., Saverno, K. R., Grizzle, A.

J. & Malone, D. C. 2011. Assessment tool for

pharmacy drug–drug interaction software. Journal of

the American Pharmacists Association, 51, 418-424.

Yu, D. T., Seger, D. L., Lasser, K. E., Karson, A. S.,

Fiskio, J. M., Seger, A. C. & Bates, D. W. 2011.

Impact of implementing alerts about medication

blackbox warnings in electronic health records.

Pharmacoepidemiology and drug safety, 20, 192-202.

Zhang, J., Xie, J., Hou, W., Tu, X., Xu, J., Song, F.,

Wang, Z. & Lu, Z. 2012. Mapping the Knowledge

Structure of Research on Patient Adherence:

Knowledge Domain Visualization Based Co-Word

Analysis and Social Network Analysis. PLoS ONE, 7.

HEALTHINF2014-InternationalConferenceonHealthInformatics