INFORMATION SYSTEMS HETEROGENEITY

AND INTEROPERABILITY INSIDE HOSPITALS

A Survey

Lucas Ribeiro

1,2

, João Paulo Cunha

and Ricardo Cruz-Correia

1,4

Center for Research in Health Technologies and Information Systems (CINTESIS)

Faculdade de Medicina da Universidade do Porto (FMUP), Porto, Portugal

Centro Hospitalar de Trás-os-Montes e Alto Douro, Portugal

Instituto de Engenharia Electrónica e Telemática de Aveiro (IEETA), Universidade de Aveiro, Aveiro, Portugal

Serviço de Bioestatística e Informática Médica da FMUP, Porto, Portugal

Keywords: Electronic Health Record, Integration, Information System Interoperability, Heterogeneity.

Abstract: This paper presents a study of the heterogeneity and interoperability of Information Systems (IS) that exist

in Northern Portuguese Hospitals. Structured interviews to each Hospital CIO were performed regarding

their opinions, existing IS and integrations. The number of IS is exponentially related of the dimension of

the hospital (number of beds), and the relation between the number of IS and the number of integrations

follows an exponential model. The ratio between the number of effective integrations and the total

possibilities is very low and follows a quadratic model, indicating that the energy spent grows rapidly with

the increase of IS number and with poor results. Admission/discharge/transfer and drug related IS are

installed more often, and therefore are better candidates for a regional network. Despite numerous efforts to

develop standards, it seems that there is a large gap between their development and their applicability.

Interoperability inside organizations is a crucial first step, looking for the goal of achieving regional and

national EHR. Standardization is vital considering the number of IS and multitude of organizations

involved.

1 INTRODUCTION

It is currently agreed that interoperability in

healthcare is a matter of great importance, but also

of great complexity. The major challenge is to find a

way to allow interoperability between different

Information Systems (IS) in order to share

information and resources. Interoperability promises

a positive effect in the quality of care and also

economic advantages (Maldonado et al., 2003, Lenz

and Reichert, 2005, Walker et al., 2005, Uslu and

Stausberg, 2008).

Several solutions coexist with the aim of

facilitating the integration of a growing volume of

data, using different standards and technologies.

After many years of development in IS, the majority

of the healthcare organizations unfortunately are still

far beyond achieving open architectures that would

allow harmonious integration of computer

applications. The creation of a cohesive and

integrated Electronic Healthcare Record (EHR) is a

more complicated task than initially expected with

several obstacles (Zviran et al., 1998, Kitsiou et al.,

2006). Consistently combining data from different

sources takes a great deal of effort because the

individual systems usually differ in several aspects,

such as semantics, data representation, functionality,

presentation and terminology (Lenz, Blaser et al.

1999; Lenz and Kuhn 2002; Kitsiou, Manthou et al.

2006). In addition, several architectural mismatches

exists in the majority of the organizations, bringing

additional problems (Land and Crnkovic, 2003).

Interoperability of EHR is on the agenda of the

European Union (e.g. EPSOS and Calliope), of

many national governments (e.g. MedCom at

Denmark, RSEpt at Portugal) and regional initiatives

(e.g. RTS at the region of Aveiro in Portugal

(Cunha, 2007)). A summary of relevant patient data

has been seen as the most appropriate approach for

establishing EHR interoperability (Shabtai et al.,

2007). Nevertheless, to obtain the full advantages of

information sharing (either for patient care,

337

Ribeiro L., Cunha J. and Cruz-Correia R. (2010).

INFORMATION SYSTEMS HETEROGENEITY AND INTEROPERABILITY INSIDE HOSPITALS - A Survey.

In Proceedings of the Third International Conference on Health Informatics, pages 337-343

DOI: 10.5220/0002749003370343

 SciTePress

management or research), it is necessary to share

much more data than just a summary. The European

Union in the January 2009 report says that achieving

full interoperability across the entirety of healthcare

would be a lengthy, expensive and possibly

unattainable goal (Veli N. Stroetmann (Ed.) and Karl

A. Stroetmann, 2009).

It is also known, that in order to obtain a regional

or national patient record it is first crucial to achieve

interoperability inside each healthcare organization

(Cotter, 2007). The integration maturity model

proposed by Schmidt includes four levels of

integration (inexistent, point to point, processes and

external); it should be noticed that only in this final

step is interoperability among different institutions

considered (Schmidt, 2000).

This paper describes the heterogeneity and

interoperability of IS existing in Northern

Portuguese Hospitals, aiming to foresee the

integration difficulties in creating a complete

regional or national patient record.

2 METHODS

2.1 Study Participants

The participants of this study are all hospitals in the

northern region of Portugal (n=34). The hospitals are

represented in this study by their Chief Information

Officer (CIO).

In the scope of this study we have just

considered clinical IS. We have not considered

integrations with any medical device.

2.2 Design Study

This is a cross-sectional study representing the

reality found in these Portuguese hospitals in the

first semester of 2009.

2.3 Data Collection Methods

ARS Norte (Administração Regional de Saúde do

Norte – Northern Regional Health Administration)

was asked permission to facilitate data collection for

the study. After their support it was sent a request

for cooperation to each Hospital. Then, the authors

sent an email to each hospital CIO involved in the

study with the questionnaire attached.

After the CIO returned the questionnaire a

process of double validation of the results was

performed: (1) the first author of this work, made a

phone interview with CIO to confirm the answers

given; (2) then the data was sent by e-mail to every

CIO so that the results of each interview were

validated by them.

2.4 Variable Description

The main variables of the questionnaire can be

grouped into CIOs opinions and the reality existing

inside their institutions. Variables about opinions:

 How important is interoperability between IS

 What plans for interoperability exist for your

institution

 Do you trust the security of the already

existing integrations regarding information

confidentiality, integrity and availability

 Are there monitoring mechanisms to detect

integration errors

 Does the hospital have a plan regarding IS

 If a plan exists, how does it refer to

interoperability issues

 Is it better to buy most of IS to one single

vendor, or to buy best-of-breed departmental

IS from multiple vendors

 Should there be regulation and certification to

the IS market regarding integration issues

 Between which stakeholders (e.g. different

hospital departments, hospital and social

institutions, hospital and primary care) should

data exchange be a priority

Variables about reality inside hospitals:

 What IS exist in the hospital

 For each IS:

 What is the scope (Global or Departmental;

the classification of the departmental

systems was made based on the list of

medical specialties of the National Board of

Physicians);

 Supplier;

 Relational database management system

(RDBMS);

 Compliance with Architecture Standards.

Classification Adapted from (Blobel, 2006)

 Terminology and Ontology Standards.

Classification Adapted from (Blobel, 2006)

 What integrations exist between IS

 For each integration:

I1. The level of the integration. Classification

adapted from (Schmidt 2000)

I2. Type of integration;

I3. Type of error detection (none, log, log and

automatic alert);

I4. Communication Standards. Classification

Adapted from (Blobel, 2006).

HEALTHINF 2010 - International Conference on Health Informatics

338

Four matrixes were built per hospital (one for each

variable regarding each integration, from I1 to I4).

Data from all arrays were stored in a relational

database, built on MS Access. Finally, we grouped

all the matrixes of all hospitals. An example of such

matrixes is illustrated in Table 1. Then some data

was exported to SPSS for statistical analysis.

Table 1: Example of possible integration matrix between 5

different systems (A to E) regarding variable I1. The letter

“d” means integration at the data level; “p” means

integration at the presentation level and “l” integration at

the logical level.

A B C D E

B d

C l d

D - - -

E d - p -

3 RESULTS

In this region there are 34 hospitals, most of them

are grouped in centers with financial autonomy (24

hospitals are grouped in 9 centers). As for the others,

7 are alone and 3 are grouped in 2 local units of care

(local units of care include hospitals and primary

care centers). The 7 ungrouped hospitals are either

very small or specialized hospitals. To our analysis

the number of different institutions considered is 18

(9 centers + 7 individual hospitals + 2 local units of

care). The number of beds per hospital varies

between 50 and 1083, with a median of 428 beds.

3.1 CIO Opinions

All CIOs (n=18) believe that interoperability is a big

issue in the future (72.2% of the responsible state it

is a very important issue and of urgent resolution,

responding the remaining 27.8% this is a very

important issue but not priority at the moment).

Most responsible for the IS have answered they

already have projects being implemented or plans to

soon begin interventions in IS integration (66.6%).

Most CIO (61.1%) stated that the hospital has a

plan for IS. The percentage of respondents who said

that the plan addresses the integration of IS in a

superficial way is 54.5% other 45.5% respond that

the issue is discussed in detail, with definition of the

requirements for interoperability.

The vast majority of respondents (72.2%)

believe that regarding the architecture of hospital IS

the best policy is to have multiple suppliers even

with the need for integrating them. The remaining

27.8% of respondents argue that the best policy is to

have a single provider that includes all clinical areas

thus ensuring a simple integration between all

modules. As for those responsible for hospitals with

500 beds or more none advocate that the best policy

is to have a single supplier.

The vast majority of CIO’s (88.9%) believe that

there should be regulation and certification of

software because without that interoperability will

hardly be effective, though the organizations should

be autonomous in the choices they make. The

percentage of those who argues that the Ministry of

Health should impose the solutions to organizations

is only 11.1%.

The three most frequently mentioned reasons, by

the eighteen respondents, to justify the lack of

interoperability that exists in the IS are: (1) existing

solutions are obsolete (66.7%), (2) the services tend

to behave as "islands" (61.1%) and (3) IS

architecture is poorly defined (50.0%).

Table 2 presents the results given, when asked to

choose from three options, as for between which

stakeholders should data exchange be a priority,

Although most respondents believe that the

integration between different IS is sufficiently

secure, they expressed some distrust in the three

areas under review (confidentiality of information

exchanged, data integrity and availability).

Table 2: Opinions of CIOs (n=18) regarding the question

about between which stakeholders should data exchange

be a priority.

Type of data exchange N %

Between each hospital department 18 100

Between hospital and primary care 18 100

Between hospital and patient 8 44

Between hospital and pharmacies 5 28

Between hospital and social security 4 22

Between hospital and private hospitals 1 6

Between hospital and insurance companies 0 0

Between hospital and patient transportation

companies

0 0

3.2 Reality Inside Hospitals

3.2.1 Existing Information Systems

We found a total of 416 different installations

resulting in 127 different IS (ratio 3.3). There are

organizations that sometimes have the same IS

installed more than once, usually as a result of the

creation of Hospital Health Centres, thus without

complete consolidation of all its IS to date. There are

three organizations in this situation (40 installations).

INFORMATION SYSTEMS HETEROGENEITY AND INTEROPERABILITY INSIDE HOSPITALS - A Survey

339

Table 3 shows the number of installations, number

of different existing systems and their ratio grouped

by the types of installed IS.

Table 3: Type of installed IS, number of installations,

number of different existing systems and their ratio. The

Departmental type IS are subdivided in subgroups.

Type of IS Installations Different IS Ratio

N % N %

ADT or EPR 67 16 8 8.4

Departmental 349 84 119 2.9

Laboratory IS 58 14 22 2.6

Imaging (RIS

and PACS)

41 10 17 2.4

Prescribing and

dispensing drugs

34 8 6 5.7

Others 216 52 74 2.9

Total 416 127 3.3

In terms of distribution of suppliers of IS by country

of origin, the country with the highest expression is

Portugal (58.3%). For the other supplier countries, it

is also worth mentioning, Spain (12.5%) and the

United States of America (10.4%), other five

countries represent the remaining 18.8%. For the

RDBMS used, the vast majority of facilities use

Oracle (77.9%) followed by SQL Server (10.3%)

and in 11.8% others.

The number of IS by organization follow a

normal distribution. The majority of organizations

(66.7%) have up to twenty two distinct IS. The

average of IS per Hospital is 20.9 (SD 9.1) with a

minimum 7 and maximum 41. If we consider the

total number of installations (including repetitions)

the average hospital installations rises up to 23.1.

No IS found follows any standard of information

architectures.

As for terminologies, 67.7% of IS don’t use any

specific terminology, LOINC is used by 11.02%;

ICD by 9.45%; SNOMED by 7.87% and others in

3.94%.

The analysis of the relationship between the

variables: number of IS (IS Nº) and number of beds

(Nº Beds); number of IS and number of effective

integrations (

Nº Integrations) and number of possible

(Pi) and effective integrations, was tested in three

models (linear, quadratic an exponential) to search

for the model that best fit the data. The quality of the

adjustment was analyzed with the F test.

There is a high positive association between the

number of beds and the number of IS (R

= 0.65).

The exponential equation better relates the number

of IS with the number of beds. The equation is as

follows:

IS Nº = 10.78

0.001 x Nº Beds

3.2.2 Existing Integrations

In the total 18 organizations (34 hospitals) we found

629 integrations. The number of integrations per

organization does not follow a normal distribution.

The median is 27.5 (minimum 7 and maximum 88

integrations).

There is a high positive association between the

number of IS and the number of integrations (R

0.66). In this case, 66% of the variation in the

number of integrations (Nº integrations) is explained

by the number of IS (IS Nº). The equation that better

represents the relationship is exponential. The

equation is as follows:

Nº Integrations = 9.15× e

0.06 x IS Nº

The total possibilities of integration (Pi), based on

the number of existing systems (IS Nº), is as

follows:

Pi= IS Nº [(IS Nº/2)-(1/2)]

The mean percentage of effective integrations

(existing integrations) versus the number of total

possible integrations is 15.8% (SD 7.6).

The graphic (scattered plot Figure 1) shows the

results regarding the number of effective

integrations and total possible integrations using the

3 models (linear, exponential and polynomial).

After analyzing the three models, we conclude

that the quadratic is the more adequate. The equation

that demonstrates this relation is as follows:

Nº Integrations=16,59+0,02 × Nº Pi+(7 × 10

-5

) × Nº Pi

The number of integrations for application layer

(data, logic, presentation) is as follows: data

(83.8%), presentation (14.9%) and logic (1.3%). In

the data layer most common integrations types are:

DB Link (35.0%) and Shared Database (30%). The

use of FTP is used asynchronously at 4.1% of cases.

In only 3.5% of cases messages are used via sockets

directly by applications. Message Oriented

Middleware (MOM) is used in 11% of cases

(Biztalk 60%; Iguana 14.3%; Mirth 7.1%; Ensemble

5.7%; Apache Service Mix 4.3%, Merge 4.3%,

Mitra 2.9% and Hermes 1.4%).

When MOM is used, the same IS uses the same

MOM in the totality of the cases. In some cases in

the same single integration is used more than one

MOM.

In the logic layer, Web Services is the only

method used. In the presentation layer all

integrations are Web based.

In the majority of the integrations no semantic

protocol is used (87.1%). In the remaining (12.9%)

the only used is HL7 V2.X. Where HL7 is used, in

80.2% cases, MOM is used in the integration.

HEALTHINF 2010 - International Conference on Health Informatics

340

Figure 1: Number of effective integrations versus number

of total possible integrations.

In most cases there is no mechanism for detecting

errors (51.5%). In 33.7% there is only log record,

and in only 14.8% of cases there are record and

alarm mechanisms for abnormal situations.

4 DISCUSSION

Numerous authors point to interoperability as crucial

for achieving gains in health, by improving the

quality of care, allowing greater convenience, cost

reduction, improvements in research, public health

and decision support, among many others. Our

results show that the CIO's of the studied hospitals

somehow support these points of view as they

attribute great importance to interoperability.

The vast majority of CIO's believe that it is not

possible to build an EHR using just a single supplier,

and so there is the need of interoperability between

existing solutions. In our opinion, this reinforces the

need for governmental regulation agencies to

facilitate the interoperability between stakeholders.

Our study concludes that there is a multiplicity of

IS, probably too many. Some are very specialized

and so very difficult to become part of a single

monolithic IS. In addition to the hospital wide IS,

the most prevalent are those related to the

laboratories, medical imaging, and drugs prescribing

and dispensing.

The number of different IS versus the number of

installations show us that global systems have the

best ratio (8.4 installations per IS), followed by drug

related IS (5.7 installations per IS). In our opinion

these are the best candidates for early multi-

institution integrations as the existing heterogeneity

is lower. The remaining IS types (Laboratory,

Imaging and others) have very similar ratios (2.6,

2.4 and 2.9 respectively) and so are more difficult to

be integrated as more different IS have to be

included.

The number of IS per organization is closely

related to the hospital dimension (number of beds),

following an exponential relationship.

The number of integrations per hospital is

closely related to the number of IS, following an

exponential relationship.

A high ratio between the real and the possible

number of integrations is more easily achieved in

small hospitals. In larger hospitals this is more

difficult to achieve since the number of IS is higher,

and although the number of integrations rise, the

total proportion is lower, because the number of

possible integrations grows faster. This means that

the energy to integrate the existing systems rises

very fast when the dimension of the hospital rises

with poor results.

We have witnessed in the last 15 years a change

of paradigm of IS "data-aware" to "process-aware”,

where organizations are increasingly focusing on the

business processes (Van der Aalst et al., 2007). SOA

and BPM promise making systems flexible and

agile. For this reasons, SOA is the “state of the art”

technology in the sector (Kuhn et al., 2007). The

adoption of SOA in health is the natural course, this

new approach will force us to rethink architecture

and how we developed the IS for the health sector

(Mykkänen et al., 2007). However, our results show

that almost all the integrations are in the data and

presentation layer, not sharing functionalities and

not taking advantage of the potential of SOA and

BPM technologies. This can possibly be explained

either by the age of these technologies, younger than

the age of the existing IS, or by the resistance to

change that sometimes characterizes the health

sector.

Despite numerous efforts to develop standards, it

seems that there is a large gap between their

development and their application in these hospitals.

No IS follows a standard information architecture,

and in 12.9% of cases the only standard of

communication found was HL7 v2.x. Interesting fact

is that HL7 is used almost always in association with

middleware (80.2%), leading us to conclude that the

IS that implement HL7 directly is very low. HL7

v2.x is the most widely implemented standard in the

world today (Cruz-Correia et al., 2007). However,

the fact that an IS is compliant with HL7 version 2,

does not imply direct interoperability between

applications, since this version has no precisely

defined underlying information model, allowing

vague definitions with a multitude of optional data

INFORMATION SYSTEMS HETEROGENEITY AND INTEROPERABILITY INSIDE HOSPITALS - A Survey

341

models. In this context, although there is great

flexibility, it is necessary bilateral agreements

between the parties involved to be able to achieve

interoperability. To solve this problem a new version

of HL7 (version 3) is developing based on a

reference model called Reference Information

Model (RIM) (Eichelberg et al., 2005). For this

reason, in the totality of the cases we studied it is

necessary bilateral agreements, even when HL7 is

used.

From a management perspective integration is

not valued as a global centralized activity. Our

results show that it is possible to find in the same

hospital technological overlapping approaches. Also,

the same IS uses the same MOM everywhere, and

therefore we conclude that the MOM is imposed by

the supplier, probably because is much faster and

easier to build the integrations with the same product

everywhere.

The construction of regional or national EHR,

are in the centre of attention today, by the potential

benefits involved. However, to make this possible,

first there must be local interoperability at each

health organization so that patient data can be seen

in a comprehensive way when it is accessed from

other institution. Interoperability inside hospitals is

weak, for the reasons presented above, thus

undermining the project of regional and national

EHR, as well as creating enormous obstacles within

organizations.

Our results also show that CIOs are reluctant

regarding the safety of the exchanged data, as in the

majority of cases (51.5%) there is no control

mechanism for the integration.

In the scope of this study we have just

considered clinical IS, and we have not considered

integrations with medical devices. If the analysis is

extended the complexity of the problem is even

greater.

A limitation of our study is the inexistence of

other similar studies for comparison. However it is

our impression that this reality will be identical in

many other regions.

Another important finding relates to the

difficulty on getting the data, due to the lack of

documentation regarding existing systems and their

integrations. In the majority of the cases getting the

data from the healthcare was hard. We have reasons

to believe that being the main author of the paper

also a CIO and therefore a colleague of the

interviewed has helped gathering the data.

5 CONCLUSIONS

Our study concludes that there is a multiplicity of IS,

probably too many. The number of IS per

organization is closely related to the hospital

dimension and the number of integrations is

exponentially related with the number of IS. The

energy necessary to integrate the existing systems

grows rapidly when the dimension of the hospital

increases, with poor results.

Almost all integrations are in the data and

presentation layer, not sharing functionalities and

not taking advantage of the potential of SOA and

BPM technologies

Despite numerous efforts to develop standards, it

seems that there is a large gap between their

development and their application in these hospitals.

To our knowledge, the situation in Portugal

seems even worse as not many (if any) people work

in international standards bodies. We intend to

present these results to national institutions aiming at

raising global awareness on our current situation.

Interoperability inside hospitals is weak, thus

undermining the project of regional and national

EHR, as well as creating enormous obstacles within

organizations.

ACKNOWLEDGEMENTS

The authors acknowledge the help provided by ARS

Norte and also the contribution provided by Drª

Filipa Gomes in the statistical analysis.

REFERENCES

Blobel, B. (2006) Advanced And Secure Architectural Ehr

Approaches. International Journal Of Medical

Informatics, 75, 185-190.

Cotter, C. M. (2007) Making The Case For A Clinical

Information System: The Chief Information Officer

View. Journal Of Critical Care, 22, 56-65.

Cruz-Correia, R. J., Vieira-Marques, P., Ferreira, A.,

Almeida, F., Wyatt, J. & Costa-Pereira, A. (2007)

Reviewing The Integration Of Patient Data: How

Systems Are Evolving In Practice To Meet Patient

Needs. Bmc Medical Informatics And Decision

Making, 7, 14.

Cunha, J. P. (2007) Rts Network: Improving Regional

Health Services Through Clinical Telematic Web-

Based Communication System. Ehealth 2007

Conference. Berlin.

Eichelberg, M., Aden, T., Riesmeier, J., Dogac, A. &

Laleci, G. B. (2005) A Survey And Analysis Of

HEALTHINF 2010 - International Conference on Health Informatics

342

Electronic Healthcare Record Standards. Acm

Computing Surveys (Csur), 37, 277-315.

Kitsiou, S., Manthou, V. & Vlachopoulou, M. (2006) A

Framework For The Evaluation Of Integration

Technology Approaches In Healthcare. Proceedings

Of The Ieee In Information And Communication

Technologies In Bioinformatics, 28–30.

Kuhn, K. A., Giuse, D. A., Lapao, L. & Wurst, S. H.

(2007) Expanding The Scope Of Health Information

Systems-From Hospitals To Regional Networks, To

National Infrastructures, And Beyond. Methods Inf

Med, 46, 500-2.

Land, R. & Crnkovic, I. (2003) Software Systems

Integration And Architectural Analysis-A Case Study.

Lenz, R. & Reichert, M. (2005) It Support For Healthcare

Processes. Lecture Notes In Computer Science, 3649,

354.

Maldonado, J. A., Robles, M. & Crespo, P. (2003)

Integration Of Distributed Healthcare Records:

Publishing Legacy Data As Xml Documents

Compliant With Cen/Tc251 Env13606. Computer-

Based Medical Systems, 2003. Proceedings. 16th Ieee

Symposium, 213-218.

Mykkänen, J., Riekkinen, A., Sormunen, M., Karhunen,

H. & Laitinen, P. (2007) Designing Web Services In

Health Information Systems: From Process To

Application Level. International Journal Of Medical

Informatics, 76, 89-95.

Schmidt, J. (2000) Enabling Next-Generation Enterprises.

Eai Journal, 7, 8.

Shabtai, I., Leshno, M., Blondheim, O. & Kornbluth, J.

(2007) The Value Of Information For Decision-

Making In The Healthcare Environment. Stud Health

Technol Inform, 127, 91-7.

Uslu, A. M. & Stausberg, J. (2008) Value of The

Electronic Patient Record: An Analysis of The

Literature. Journal Of Biomedical Informatics, 41,

675-682.

Van Der Aalst, W. M. P., Benatallah, B., Casati, F.,

Curbera, F. & Verbeek, E. (2007) Business Process

Management: Where Business Processes and Web

Services Meet. Data & Knowledge Engineering, 61, 1-

Veli N. Stroetmann (Ed.), D. K., Pierre Lewalle, Alan

Rector, Jean M. Rodrigues, & Karl A. Stroetmann, G.

S., Bedirhan Ustun, Martti Virtanen, Pieter E. Zanstra

(2009) Semantic Interoperability For Better Health

And Safer Healthcare. European Commission I N Fo

Rma T I O N S O C I E T Y A N D M E D I A.

Walker, J., Pan, E., Johnston, D., Adler-Milstein, J., Bates,

D. W. & Middleton, B. (2005) The Value Of Health

Care Information Exchange And Interoperability.

Health Affairs.

Zviran, M., Armoni, A. & Glezer, C. (1998) His/Bui: A

Conceptual Model For Bottom-Up Integration Of

Hospital Information Systems. Journal Of Medical

Systems, 22, 147-159.

INFORMATION SYSTEMS HETEROGENEITY AND INTEROPERABILITY INSIDE HOSPITALS - A Survey

343