HEALTH INFORMATICS

ELECTRONIC MEDICAL RECORDS

Electronic medical records

TERRY J. HANNAN

Historical overview
The importance of medical records in health care delivery has been recognised for a long
time. Its relevance to patient care and health administration was documented by Florence
Nightingale in 1873 a book entitled Notes on a Hospital. Health care is a continuous process
in which data is progressively accumulated, therefore the record must function as a pre-birth
to post-death system that meets the requirements for any clinical setting-whether intensive
care or primary care (Hammond 1993).
Ideally the medical record should be the primary repository of all information regarding
patient care, provide decision-support, and be a tool for supporting and maintaining ancillary
health care activities such as administration, quality assurance, research and epidemiology.
Shortliffe has defined medical (health) practice as medical decision-making, (Shortliffe
1990) and it is recognised that there is an integral relationship between medical decisionmaking, the accumulation of clinical data, health care costs, patient outcomes, and the quality
of care (Johns & Blum 1973, James 1989). The delivery of quality, cost-effective health care
requires efficient decision-support tools based on the medical record system if these end
points are to be achieved.

Traditional medical record systems

Current medical record systems are predominantly hard copy paper-based models with or
without variable components of electronic data such as laboratory results and X-ray reports.
The paper chart can be read by only one person at a time and they must have it physically in
their possession. It is difficult to store and retrieve, requiring space, time and effort. It can be
organised in one format at a time yet the demands of the users of the record require it to be in
a multitude of formats to meet the individuals needs. To reorganise the record into a variety
of formats requires major time commitments and the schematic format is easily corrupted.
The paper record is not always legible, is often inaccurate, lacks clinical sensibility, and is not
compatible with specified data standards or other information stored in the record (Tierney
&Hannan 1992).
Poor indexing of data makes the finding of information difficult or
impossible (Brennan et al 1991, Leape et al 1991).

1/15

1/15

Reproduction of the manual record by transcription and photocopying adds to the costs of
health care services without any corresponding proven benefit. It was recently calculated in
the U.S.A. to cost an additional $15 billion per year to manually reproduce the medical record
by transcription (Frawley 1993). Use of the paper chart as a medical record impedes efforts to
monitor and improve health care by the inherent difficulty, time, and expense required to
access individual charts(Payne 1990).

Technology and health care delivery

Since the 1960s there has been a rapid growth in the technology used to support medical care
(Blum 1986, Orthner & Blum 1989, Shortliffe 1990), and this has resulted in the creation of
enormous volumes of data and information that is available to assess and manage the
delivery of health care. Weed estimated that an individual patient can generate up to 50,000
data items during their life (Weed 1989), and to support good decision-making individuals
who provide health care require timely integration of this data.
New and evolving
technologies continue to produce and store large volumes of data and information for patient
care (Blum 1986, McDonald 1989a, McDonald 1992a), but there only a few systems that
provide information processing tools which support clinical decision making (Blum 1986,
Orthner & Blum 1989, Shortliffe1990).
It is recognised that information processing capacity of the human brain is limited in its
ability to accurately decipher this clinical data and information in a timely manner without
errors (Miller 1956, Pryor & Clayton 1991, McDonald 1976). Errors in decision making are
further increased when there is random noise data, for example, with unexpected data input
in stressful situations, a common occurrence in health care.

Consequences of using manual-based record systems

The use of predominantly manual, non-integrated medical records systems has led to
increasing costs in patient care (Johns & Blum 1973, Blum 1986) and administration
(Woolhandler & Himmelstein 1993), decreased compliance with health care standards
(McDonald 1976, Gardner & Schafner 1993), inappropriate variation in health care delivery
(Wenneberg & Gittelsohn 1973), and possible negligent behaviour by health care
providers(Brennan et al 1991, Leape et al 1991). A Harvard study into the incidence of
adverse events during hospitalisation led the authors to conclude that, Lawyers generally
believe that investigation of substandard care only begins with the medical record; that in
many instances the medical record even conceals substandard care; and that substandard care
is not reflected in, or discoverable in the medical record (Brennan et al 1991, Leape et al
1991).

Concept of electronic medical record (EMR)

What is the Electronic Medical Record (EMR)? It is the storage of all health care data and
information in electronic formats with the associated information processing and knowledge
support tools necessary for the managing the health enterprise system.
In the early 1970s several institutions investigated the concept of creating an EMR to
improve patient care. An important feature each of these projects was the concept that the

medical record should be the cornerstone for all information systems within the health care
environment, (Blum 1986, McDonald et al 1992a, Safran 1990, Enterline et al 1989,
Kuperman et al 1991) and that the data supporting ancillary patient care activities such as
administration, pharmacy, laboratories, etc., could and should be generated as a by-product of
the patient care process (Blum 1986, Slack 1990, Bleich et al 1985).
One of the earliest successful implementations of EMR functions was at the Regenstrief
Institute, Indianapolis. Using the Regenstrief Medical Record System (RMRS) McDonald
demonstrated that the use of computer-generated reminders based on patient-specific
laboratory data resulted in a reduction of physicians errors in the detection of life-threatening
events, and also confirmed that busy physicians were often unable to detect many of the
critical abnormalities occurring in the patient record. He concluded that, the amount of data
presented to the physician per unit time is more than he can process without error. The
computer augments the physicians capabilities and thereby reduces his error rate....It is very
likely that the physicians in these studies were simply unable to detect all the multitudinous
conditions specified by the standards. (McDonald 1976) Computer-generated reminders are
now used as standard tools for patient care in the RMRS and other EMR systems used in
hospitals and ambulatory care environments (McDonald et al 1992a, Safran 1990, Safran et al
1992, Enterline et al 1989, Kuperman et al 1991).

The Institute Of Medicine study into electronification of the

patient care record
In 1992 the Institute Of Medicine (Dick & Steen 1991) of the American National Academy of
Sciences, published the results of its study into computerised medical records, their
functionality, and how technology could bring the benefits of these records within the reach
of all those within the health care system. The recommendations from this study are
summarised in the following list:

Health care professionals and organizations should adopt the computer based patient
record (CPR) as the standard for medical and all other records related to patient care.
To accomplish Recommendation 1, the public and private sectors should join in
establishing a Computer based Patient Record Institute (CPRI) to promote and facilitate
development, implementation, and dissemination of the CPR.
Both the public and private sectors should expand support for the CPR and CPR system
implementation through research, development and demonstration projects. Specifically,
the committee recommends that Congress authorise and appropriate funds to implement
the research and development agenda outlines herein.
The committee further
recommends that private foundations and vendors fund programs that support and
facilitate this research and development agenda.
The CPRI should promulgate uniform national standards for data and security for
facilitate implementation of the CPR and its secondary databases.
The CPRI should review federal and state laws and regulations for the purpose of
proposing and promulgating model legislation and regulations to facilitate the
implementation and dissemination of the CPR and its secondary databases and to
streamline the CPR and CPR systems.
The costs of CPR ststems should be shared by those who benefit from the value of the
CPR. Specifically, the full costs of implementing and operating CPRs and CPR systems

should be factored into reimbursement levels of payment schedules of both public and
private sector third-party payers. In addition, users of secondary databases should support
the costs of creating such databases.
Health care professional schools and organizations should enhance educational programs
for students and practitioners in the use of computers, CPRs, and CPR systems for patient
care, education, and research.

Publication of this report has resulted in wide range of activities directed towards
standardised EMR developments. In North America, the US. Department of Health and
Human Services, has implemented a national policy on a health information communication
infrastructure based on automation of the patient record.(US. Department of Health and
Human Services) This group sees a national interconnected communication network linking
all participants in the health care system via their own computer-based patient record
system -an information system that would have the ability to create, store, retrieve, transmit
and manipulate patients health data in ways that best support decision making about their
care. These record systems would be linked to reference bases of aggregated patient data and
computerised knowledge-based systems which use decision support logic and practice
guidelines to help caregivers make better decisions about diagnosis and treatment options
(Safran et al 1990, Kuperman et al 1991).

Software applications providing decision support in EMRs

No complete EMRs currently exist however standards for software functionality and
decision-support have been defined and are seen as core elements for future EMR
developments (Dick & Steen 1991). The basic software components necessary for future
EMRs as defined by McDonald are shown in Table 12.1, and a description of these terms is
outlined in the following text (McDonald 1988).

Maintain a data dictionary

Orientation
Introspection
Selectivity of data input
Query languages

In systems which maintain a data dictionary, all data and observations are stored in
records, which include fields that link or point to the dictionary files. This means that the data
is stored in coded formats which provides for more consistent recording and ease of data
entry. They also provide facilities for declaring the data entry fields prior to the recording and
storage of data, without the need to define in advance the space to be occupied by the
recorded data, thus providing a much more economical use of computer storage space, and
more rapid access to the data.
Orientation provides the facility to produce an array of time-oriented flow sheets from the
stored data. Figure 12.1. shows summary flow chart of clinical data taken from a patient on
chemotherapy for leukaemia. The capacity for displaying clinical data in user-defined, timeoriented formats, is a decision support tool fundamental to good clinical practice (Blum 1986,
Orthner & Blum 1989, Hannan 1991).

CCCIS CLINICAL INFORMATION SYSTEM

HISTORY NO: 808080

NAME: MARROW,BONE
DATE: 30/10/88

FULL FLOW
:-----------------:-------:-------:-------:-------:-------:-------:-------:
: CURRENT
:04SEP87:05SEP87:07SEP87:08SEP87:09SEP87:10SEP87:11SEP87:
:PROTOCOL/CYCLE --:-------:-------:-------:-------:-------:-------:-------:
:525
:
:
:
:DAY
1:DAY
2:DAY
3:DAY
4:
:
CYCLE :
:
:
:C2D1
:C2D2
:C2D3
:C2D4
:
:737
:DAY 24:DAY 25:DAY 27:DAY 28:
:
:
:
:
CYCLE :C1D24 :C1D25 :C1D27 :C1D28 :
:
:
:
:CHEMO
--------:-------:-------:-------:-------:-------:-------:-------:
:ARAC
MG
:
:
:
:
220:
220:
220:
220:
:DAUNORUB MG
:
:
:
:
110:
110:
:
:
:VP-16 IV MG
:
:
:
:
165:
165:
165:
165:
:HAEM.
--------:-------:-------:-------:-------:-------:-------:-------:
:WBC
10^9/L :
4.1:
5.1:
7.8:
8.4:
:
5.7:
:
:RBC
10^12/L :
3.88:
4.15:
4.23:
4.13:
:
4.74:
:
:HGB
g/dl
:
11.7:
12.2:
12.8:
12.7:
:
14.6:
:
:HCT
%
:
33.3*
36.5*
37.8:
37.1:
:
42.7:
:
:MCV
fL
:
86.0:
88.0:
89.0:
90.0:
:
90.0:
:
:MCH
pg
:
30.2:
29.4:
30.3:
30.8:
:
30.8:
:
:MCHC
g/dl
:
35.1*
33.4:
33.9:
34.2:
:
34.2:
:
:BAND FMS %
:
:
:
2:
:
:
:
:
:NEUTROPH %
:
30*
36*
44:
45:
:
68:
:
:PLATELET 10^9/L :
67*
103*
214:
247:
:
200:
:
:BLASTS
%
:
:
7:
:
:
:
:
:
:MYEL
%
:
3:
:
:
:
:
:
:
:METAMY
%
:
2:
:
:
:
:
:
:
:EOS
%
:
:
1:
:
5:
:
:
:
:TOX.GRAN
:
SL:
:
SL:
SL:
:
:
:
:LYMPH
%
:
43.0:
46.0:
42.0:
36.0:
:
23.0*
:
:MONO
%
:
22.0*
10.0:
12.0*
14.0*
:
9.0:
:
:ANISOCYT
:
:
:
MO:
:
:
:
:
:OV.POIK
:
:
:
MO:
:
:
:
:
:POLYCHRO
:
:
:
SL:
:
:
:
:
:RNDMACRO
:
:
:
SL:
:
:
:
:
:CHEMSTRY---------:-------:-------:-------:-------:-------:-------:-------:
:SODIUM
MMOL/L :
144:
:
139:
:
:
138:
:
:POTASS
MMOL/L :
3.5:
:
4.3:
:
:
3.5:
:
:CHLOR
MMOL/L :
105:
:
103:
:
:
108:
:
:CO2
MMOL/L :
26:
:
25:
:
:
26:
:
:UREA
MMOL/L :
6.3:
:
5.5:
:
:
6.3:
:
:CREAT
MMOL/L :
0.08:
:
0.09:
:
:
0.08:
:
:GLUCOSE MMOL/L :
5.6*
:
:
:
:
:
:
:BILI.T
UMOL/L :
:
:
:
18*
:
14:
:
:AST
U/L
:
:
:
53*
:
:
34:
:
:ALT
U/L
:
:
:
131*
:
:
86*
:
:GGT
U/L
:
:
:
114*
:
:
83*
:
:ALK.PHOS U/L
:
:
:
198*
:
:
143*
:
:TOT.CALC MMOL/L :
:
:
2.26:
:
:
2.16:
:
:PHOS
MMOL/L :
:
:
1.3:
:
:
1.2:
:
:T.PROT
G/L
:
:
:
74:
:
:
66:
:
:ALBUMIN G/L
:
:
:
39:
:
:
34*
:
:-----------------:-------:-------:-------:-------:-------:-------:-------:
(C)ONTINUE OR (Q)UIT
(C)URRENT
(E)ARLIEST
(D)ATE
(B)ACKWARD
(F)ORWARD
(P)RINT
(Q)UIT

Figure 12.1 User defined flow chart of clinical data

Introspection is a decision support tool where the computer is able to examine data and
information stored within the EMR database using pre-defined clinical rules, and identify
certain conditions that require attention (McDonald et al 1992a, Kuperman et al 1991, Safran
et al 1990). These record systems automatically produce drug-alerts, warnings, protocolgenerated reminders and are able to detect significant alterations in data elements which
complement the medical decision making process. (McDonald 1976, Kuperman et al 1991)
The users of EMR systems must decide on the selectivity of data input and how it will be
entered into the medical record, manually, electronically, or by other processes (McDonald
1990a). Quality data input can be expensive to maintain because it requires disciplined, welltrained staff. (Orthner & Blum 1989, Enterline et al 1989). However, once data are stored the

users have access to the computers powerful report generator functions that provide useful
data and information displays and reports that support all components of health care delivery
(Enterline et al 1989, Safran et al 1990).
It is now possible to access these large volumes of clinical data stored in electronic
formats using medical query languages for the purposes of research, epidemiology, health
care planning, and for producing reports based on data analysis. The collection of information
on large numbers of patients to answer specific problems is expensive, time consuming and
personnel intensive and is often restricted to small numbers of patients for a limited range of
conditions (Tierney & Hannan 1992). For example, to answer the question whether the
postmenopausal oestrogen use protected women against cardiovascular morbidity and
mortality, Stampfer and others had to receive regular mailed reports on specific clinical
events from more than 48,000 women for 10 years (Stampfer 1991). These tasks could have
been performed more easily and cheaply if the data were available in a single EMR database
or accessible over a range of standardised EMR systems.

Computerised clinical decision support tools

Specific decision support tools for use within EMRs have been defined which provide
benefits to health care, and these are as follows: (Clayton & Pryor 1991).

Alerting
Interpretation
Assisting
Critiquing
Diagnosing
Management

Alerting automatically provides decision makers with data and information in situations
where rapid, sometimes life threatening, decisions are required. Examples are abnormal
laboratory values, vital sign trends, failure to perform nursing procedures and medication
contraindications. These clinical situations often have episodes of unpredictable random
noise data which impair the decision making process leading to errors in patient care.
A system of alerts is used routinely in the HELP (Health Evaluation through Logical
Processing)(Bradshaw et al 1989) and Regenstrief Medical Record Systems (RMRS)
(McDonald et al 1992a). Established benefits from the use of alerts are a reduction in
physician and nursing errors in patient management, increased compliance with predefined
standards of care, (McDonald et al 1984) decreased length of stay in hospital and time spent
in life-threatening situations (Kuperman et al 1991, Sittig et al 1989).
Using automated alerts in the HELP system during surgery for non-indicated and nonordered antibiotics Classen demonstrated a fall in the post-operative infection rate from 13%
patients per day to 5.5%, and a fall from 35% to 18% in the percentage of patients receiving
antibiotics late for surgery. As a consequence there was a reduction in the number of patients
receiving antibiotics for an excessive time post-operatively which produced overall savings of
$59,000 in 6 months (Classen 1992). This system for recommending antibiotics has now been
extended to primary and ambulatory care (Evans 1991, and 1993).

Interpretation is where stored clinical data is assimilated leading to an improved

understanding of what the data means. Examples are ECG interpretation, blood gas data
analysis and the interpretation of X-ray findings (Blum 1986, Orthner & Blum 1989,
McDonald 1989, 1990b, 1991).
Assisting is where the use of decision-support tools speed up or simplify some clinical
action. This technique is used in the production of clinical orders, nursing assessment of
patients, and history and physical examination. Patient pre-printed encounter forms improve
history taking and standardise data recording. Assisting also facilitates direct data entry onto
computer terminals making the data immediately available to authorised users of the EMR
(McDonald et al 1992a).
Where assisting is used in the ordering of blood samples the specimen is ordered online and the EMR system indicates the tube type, laboratory to which the specimens are to be
sent and when the last specimen was ordered. In this situation the system may also
recommend how many times and on what dates a given sample is to be collected within a
specified time interval (Enterline et al 1989, McDonald et al 1992a). At the Johns Hopkins
Oncology Centre (JHOC), computerised protocol-directed care plans were used to manage
blood-product facilities by recommending specific numbers of platelet units for
thrombocytopenic patients who were at risk of bleeding. This produced cost savings of
$250,000 per year, decreased the use of a limited resource (platelets), and improved patient
outcomes (no increase in bleeding) (Enterline et al 1989).
Critiquing is where decisions are analysed within the EMR using defined knowledge
rules to verify the appropriateness of those decisions. The system is able to recommends to
the physician, nurse, etc., the most appropriate decision to make.(McDonald 1976) Examples
of critiquing include clinical orders, protocol-directed care plans, diagnosis making and
management plans (Blum 1986, Orthner & Blum 1989).
Diagnosing is where a specific clinical model is applied for the purpose of understanding
a complex clinical situation. In these situations the computer may provide a probability list
for a range of differential clinical diagnoses based on the data has stored within it (Miller et al
1982, Safran et al 1991). Currently these systems are limited to small clinical domains, such
as intensive care, and are expensive because of the expertise needed to maintain the
knowledge rules within them (Blum 1986, Orthner & Blum 1989).
Management is the generation of action oriented decisions designed to improve the
functionality of the current system state. Examples include hospital operations, resource
allocation (including personnel) and the current status of changing clinical disease patterns,
either acute or chronic. In HELP the decision support system will recommend changes in
FI02 (Fraction of Inspired Oxygen) with patients on respiratory support and suggest when to
draw the next blood gases based on existing laboratory and clinical data (Kuperman et al
1991). Pooled data from HELP made available to surgeons performing uncomplicated
prostatic resections resulted in reduced length of stay and costs of the procedure over a range
of hospitals in the region (James 1989, Grandia 1994).
The Oncology Center Information System (OCIS) system uses nurse generated patient
dependency ratings to allocate staff during hospital admissions, and patients are scheduled in
outpatient clinics according to the procedures performed so that the doctor-patient encounters
coincide with the availability of the clinical data.

Many decision-support applications coexist within EMR systems and they must be
integrated to the continuously expanding database of individual patients and groups of
patients. Data required for patient care must be available in a timely, reliable and complete
manner with the user being able to extract data they require in the format that best suits their
decision making (Blum 1986, Kuperman et al 1991). Timeliness of data retrieval is critical in
patient care and it has been recommended that data recall times of less than two seconds
should be achieved irrespective of the complexity of the decision support function (Clayton &
Pryor 1991). An example of how access, storage, and manipulation of clinical data is able to
assist all levels of health care is illustrated in Fig.12.2:
CCCIS CLINICAL INFORMATION SYSTEM
PRINCE OF WALES/PRINCE HENRY HOSPITALS
888888: WALKER, LUKE HEMATOLOGY PLOT FOR CYTOPENIC PATIENTS 03/11/89
WBC,PLATELETS,HGB,TEMP FOR CYTOPENIC PATIENTS
:---------------------------------------------------------:
10.0W:
:
:
:
1000P:
:
8W:W
wW :
500P: w
W w :
: w
w
w:
250P: w
w pPppw:
: w
w pP
P:
125P: w
wWpp
:
: w
wWpP
:
4W: Pp
wpP
:
50.0P:----P---P------------------------------------------------:
:P w P
Pppp
wWp
:
31.25P:
Ww
Pp pppP p
pP
wPwWpP
:
1.0W:------W-W---------P-----P-----------P---P----------------:
15.625P:
WwwwWwWwwwWwWwPwwwWwPwP
:
0W:
:
:---------------------------------------------------------:
ARAC
: A A A A A A A
A A A:
DAUNORUB: D D D
D D :
VP-16 IV: V V V V V V V
V V V:
:---------------------------------------------------------:
1987
AUG
SEP
DATE
12 14 16 18 20 22 24 26 28 30 01 03 05 07 09 11 13
(N)EXT (B)ACK (F)ORWARD (D)ATE (E)VENT (R)EDO (P)RINT (Q)UIT
Figure12.2 Daily plot of white blood cells and platelets with chemotherapy

This is a plot of white blood cells and platelets on a patient undergoing chemotherapy.
The display shows user-defined levels (horizontal dashed lines) at which decisions relating to
platelet or white cell transfusion, or prophylactic antibiotic therapy may be made. Using
accumulated data from groups of patients based on the time interval in which they have a low
cell counts, the length of stay in hospital can be predicted by treatment and diagnosis, thus
aiding bed allocation and rostering of nursing staff based on patient dependency status.
Resource utilisation such as the number of tests ordered, medications utilised and bed
occupancy can be evaluated from the data accumulated during the patient care process,
demonstrating how administrative and health planning data can be generated from the patient
care record. The data displayed is linked to the complete medical record so complex
interrelationships of data within the EMR can also be evaluated (Bleich 1985, Slack 1990,
Safran et al 1990, and 1991, Grandia 1994).

Existing EMR models and confirmed benefits of EMRs

Complete automation of the medical record system has not yet occurred, however there are a
range of EMR systems which provide information demonstrating the real and potential
benefits of electronically stored medical information. Recent advances in technology relating
to data and information storage, such as Compact DisksRead Only Memory (CD-ROM),
provide facilities for life-long repositories of medical data. There are functioning EMR
systems storing in excess of 1 million patients on line, representing billions of data points
accessible within seconds (Safran 1990, McDonald 1989a), and the ability to manage these
large volumes of data is another area where the benefits of computerisation can be seen. For
example;

Fries and his co-workers, using the American Rheumatological Association Medical
Information System (ARAMIS), showed that physicians using a computerised data flow
sheet were able to find specified information twice as fast as physicians using the
traditional paper record (Fries 1974). The physicians in this study were able to find the
relevant information virtually all the time with the computer record but failed to find 10%
of the information in the paper chart,
Whiting-OKeefe and his colleagues used the Summary Time-Orientated Record (STOR)
to show that physicians were better able to predict their patients laboratory results when
using the computerised record (Whiting-OKeefe 1980).
Computer-generated reminders used in the RMRS have improved compliance with
practice standards particularly in the area of preventive care (McDonald et al 1984,
1989b, and 1990c, Litzelman et al 1990).
HELP has an extensive alerting system based on data generated from laboratory auto
analysers which has resulted in shorter length of stay in hospital, reduced costs and time
spent in life-threatening situations (Kuperman et al 1991).
Research using OCIS has confirmed the benefits of protocol-directed care plans by
reducing resource utilisation and costs without reducing the quality of care (Enterline et al
1989).
Use of the Beth Israel Hospital (BIH) system has improved access to medical records,
provided decision support through rapid access to bibliographic databases and to
cumulative patient data subsets via its medical query language (Safran et al 1990, and
1991).

The benefits from effective automation of health care delivery using EMR systems based
on the IOM criteria can be viewed from different domains namely, patient care, quality
assurance, costs, epidemiology, research, and administration.

Patient care
Preventive care is recognised as one area for providing major cost savings in health care
delivery. The effective use of an electronic reminder system based on the patient record has
resulted in the more effective implementation of preventive care protocols, immunisation
procedures, better utilisation of vaccines, and a reduction in morbidity and mortality from
infectious diseases (Fries et al 1993, Gardner & Schafner 1993).

Quality assurance
Effective quality assurance in health care requires a reduction of inappropriate variation in
health care delivery and adequate documentation of procedures to improve the processes
involved in health care delivery (Wennberg & Gittlesohn 1973, James 1989).
At the
Regenstrief Institute rheumatologists use the RMRS to identify patients with rheumatoid
arthritis and other connective tissue disorders who are being treated with second line drugs
such as gold, methotrexate, azathioprine, and penicillamine. The monitoring program detects
whether the appropriate pre-treatment haematology and urine tests have been performed, if
the results are abnormal and whether the drugs were held or given in reduced dosages.
Patients whose records do not meet these standards have their outpatients charts reviewed
and, if substandard care has been delivered, the providers are contacted (McDonald 1990c).
The same system using data accumulated within the first 24 hours following admission, has
been used to identify patients who are likely to have high hospital costs (McDonald et al
1990c). Tierney and others (Tierney et al 1993) demonstrated how the use of an electronic
medical record system to control orders for inpatients significantly lowered patient charges
and hospital costs. In this situation the ordering of blood tests, medications, etc, was
performed at the computer terminal by the medical officers and the justification for requesting
each test was prompted by reminders from the EMR.

Research
Research costs represent a significant component of the health care budget as the acquisition
of accurate data is difficult and costly in both time and dollars (Tierney & McDonald 1991)
Payne demonstrated the cost and time savings benefits of the electronic medical record
system Computer STored Ambulatory Record (COSTAR) over the paper record when
evaluating the effects of anti arthritic medications in hypertensive patients. The costs saving
were in the tens of thousands of dollars and the time savings were measured in hundreds of
person hours (Payne 1990).

Epidemiology
As chronic diseases become more prominent in modern health care, larger patient populations
will need to be studied to detect variations in diseases and changes in health outcomes
resulting from therapeutic interventions, and the size of these populations make it impractical
to undertake studies using existing paper-based medical record systems (Tierney et al 1985).
Effective studies of these large patient populations will require new analytical methodologies
to be created so we can harness all the information they contain (McDonald 1989b).
EMR systems have been used to identify those patients with hypertension prescribed nonsteroidal anti-inflammatory drugs who are likely to develop renal insufficiency (Tiernery et al
1989, Murray et al 1990), and identify clinical factors that predict which patients taking
diuretics are likely to develop hypokalemia (Tierney et al 1985a,) and ventricular ectopy
(Tierney et al 1985b).

Administration
Data stored within the HELP clinical system has been used in the Utah Intermountain Health
Centre to evaluate the costs and quality of care for a variety of medical and surgical
procedures. The EMR provided information on variations in length of stay, surgical procedure
time and costs for uncomplicated surgical resections of the prostate. This information was
relayed to the medical practitioners who performed the operations in a range of institutions
and they co-operated in altering their surgical techniques and procedures to reduce the mean
length of stay and costs of routine Transurethral Resections of the Prostate (TURP) (James
1989). In the evaluation process it was found that certain preoperative procedures such as
chest X-rays, were unnecessary if a patient had an uncomplicated medical history. The same
clinical data generated from the patient care process is being used to measure and evaluate the
effectiveness of Diagnostic Related Group (DRG) studies, and has led to a reduction in DRG
costings. Similar hospitals not using a patient-based EMR system have expanding DRG and
other health costs (Gardia 1994). This confirms Howard Bleichs observation that 90% of
administrative cost data can be generated as a by-product of the patient care process (Bleich
1985).
To achieve these results EMR systems must provide standardisation of data recording so
that data and information can be shared across institutions and internationally. Achieving the
end points of EMR integration is not easy, however successful implementations have been
completed and they provide answers to many of the difficulties faced and how they can be
overcome (Martin 1992, Hannan 1994).

Conclusions
With the knowledge that established models of EMRs have been shown to improve the health
care process, what is the recommended course for future EMR projects?
Developers, project directors and users of clinical information systems must begin to
share the experiences, and use existing and evolving software tools to reduce implementation
costs. This will result in the most effective delivery quality health care.
References
Bleich H L, Beckley R F, Horowitz G L, et al and others. Clinical computing in a teaching
hospital. N Engl J Med. 1985; 312: 756-64
Blum B I Clinical Information Systems. Johns Hopkins University:Springer-Verlag, New
York; 1986; ISBN: 0-387-96190-9
Brennan T A, Leape L L, Laird N M, et al Incidence of adverse events and negligence in
hospitalized patients. Results of the Harvard Medical Practice Study I N Engl J Med 1991 Jul
18;325(3):210
Enterline J P, Lenhard R E Jr, Blum B I A Clinical Information System for Oncology. New
York: Springer- Verlag, New York.; 1989; ISBN: 0-387-96956-X

Bradshaw K E, Gardner RM, Pryor TA Development of a Computerised Laboratory Alerting

System. Comput. Biomed. Res. 22:575-587;1989
Classen D C, Evans S, Pestotnik S L, Horn S D, Menlove R L, Burke J P, The
timing of propyhlactic administration of antibiotics and the risk of surgical wound infection.
N Engl J Med. 1992; 326: 281-6
Dick R S, Steen E B The Computer-based patient record. An essential technology for health
care. Washington DC 20418: National Academy Press; 1991; ISBN: 0- 309-04495-2
Evans R S The HELP system: A review of clinical applications in infectious diseases and
antibiotic use. MD Comput 1991; 8: 282-8
Evans R S, Pestotnik S l, Classen D C, et al Development of an automated antibiotic
consultant. MD Comput 1993; 10:17- 22
Fitzgerald J F, Tierney W M, Engeler M T, et al Predicting inpatient charges. A model using
clinical admission information. Clin Res. 1990; 38: 912A
Frawley K A Health Information: overview of privacy and confidentiality issues. Tutorial 3
AMIA Spring Congress St Louis, Missouri. May 9-12 1993
Fries J F Alternatives in medical record formats.Med Care 1974;12:871-81
Fries J.F., Koop C.E., Beadle C.E. et al Reducing health care costs by reducing the need and
demand for medical services N.Engl J Med 329;321-5; 1993
Gardner P, Schaffner W Immunisation in adults. N Engl J Med 328;1252-8;1993
Gardner R M, Golubjatnikov OK, Laib RM, et al Computer-critiqued blood ordering using
the HELP system. Computers and Biomedical Research. 1990 Dec; 23(6): 514-528
Grandia L D Intermountain Health Cares Medical Informatics Informatics Strategy. Plenary
Session AMIA Spring Congress, San Francisco May 5 1994
Hannan T Quality assurance. Aust. Clin. Rev.. 1991; 11: 22- 7
Hannan T Internationak transfer of the Johns Hopkins Oncology Center Information System
MD Computing 1994 11:2; 92-99
Hammond W Ed. An in depth look at the data structure of a computer-based patient record TMR. AMIA Spring Congress, St Louis, Missouri. May 9-12 1993. p43
James B C Quality management for health care delivery.The hospital research and
educational trust, 840 North Lake Shore Drive Chicago, Illinois 60611.1989 ISBN 0- 87258537-9

Johns R J, Blum B I The use of clinical information systems to control costs as well as
improve patient care. Trans Am Clin Climatol Assoc. 1973; 90: 140-152
Kuperman G J, Gardner R M, Pryor T A HELP: a dynamic hospital information system.
Springer-Verlag, New York. 1991; ISBN 0-387-97431-8
Leape LL, Brennan TA, Laird N, et al The nature of adverse events in hospitalized patients.
Results of the Harvard Medical Practice Study II. N Engl J Med 1991 324:377-84
Litzelman, DK; Dittus, RS; Miller, ME; Tierney, WM. Improving compliance by forced
choice preventive care reminders. Clinical Research. 1990; 38.
Martin D K Making the connection:the VA-Regenstrief project. MD-Comput.1992;9: 91-6
McDonald C J Protocol-based computer reminders, the quality of care and the nonperfectibility of man. NEJM. 1976; 319(18):1351-1355
McDonald C J, Hui S L, Smith D M, et al, Reminders to physicians from an
introspective computer medical record: A two year randomized trial. Ann Intern Med.
1984;100:130-38
McDonald C J, Hui S L, Tierney W M Diuretic-induced laboratory abnormalities that predict
ventricular ectopy. J Chron Dis. 1986; 39: 127-35.
McDonald C J Computer-Stored Medical Record Systems. MD Comput. 1988; 5(5):4-5
McDonald C J Medical Information Systems of the Future. MD-Computing. 1989; 6(2): 8287
McDonald C J Risk versus Benefits of Tests and Treatments. MD-Computing. 1989; 6(2):
63-64
McDonald C J Input Technology. MD-Computing. 1990; 7(4):201-204.
McDonald C J Interchange Standards Revisited. MD Comput.1990; 7(2):72-74.
McDonald C J Analysis of complex data,image reconstruction, and Joshua Willard Gibbs.
MD-Comput.1991;8:272-4.
McDonald C J, Tierney W M, Overhage M, et al, The Regenstrief Medical Record System:
20 years experience in hospitals, clinics, and neighbourhood health centers. MD Comput.
1992; 9(4):206-17.
McDonald C J Delivering X-ray images on hospital computer networks. MD-Computing.
1992;9:348-50.
McDonald C J Hui S L Tierney W M, Effects of computer reminders for influenza
vaccination on morbidity during influenza epidemics. MD-Comput.1992;9:304-12.
McDonald C J, Tierney W M, Miller M E, et al, Toward qualification of small area
variations. Clin Res. 1990; 38: 914.

Miller G A The magical number seven plus or minus two: some limits on our capacity for
processing information. Psychol Rev 1956 63:81-97.
Miller RA, Pople HE Jr, Myers JD, INTERNIST-1: an experienced computer-based
diagnostic consultant for general internal medicine. N Engl J Med 307:468 1982
Murray M D, Brater D C, Tierney W M et al, McDonald C J. Ibobrufen-associated renal
dysfunction in a large general internal medicine practice. Am J Med Sci 1990; 299: 222-9
Orthner H O, Blum B I, Implementing Health Care Information Systems. Springer-Verlag,
New York.1989; ISBN 0-387-96690-0
Payne T H, Goroll A H, Morgan M, et al, Conducting a matched-pairs historical cohort study
with a Computer-Based Ambulatory Medical Record System. Comput. Biomed.Res. 1990;
23(5): 455-472.
th

Pryor T A, Clayton P D, Decision support systems for clinical medicine. Tutorial II 15

Annual Symposium on Computer Applications in Medical Care. Washington DC. November
17 1991.

Safran C, Porter D, Rury, CD, et al, Clinquery:Searching a Large Clinical Database. MDComput. 1990; 7(3): 144-153.
Safran C, Rury C, Rind D M, et al, A computer-based outpatient medical
record for a teaching hospital. MD-Computing. 1991; 8: 291-9.
Shortliffe
E. H.; Perreault, L. E. Medical Informatics. Massachusetts: AddisonWesley,Massachusetts; 1990; ISBN: 0-201- 06741-2.
Sittig D F, Nathan L P, Gardner R M, et al, Implementation of a computerised patient advice
system using the HELP clinical information system. Comput.Biomed. Res. 1989;22:474-487.
Slack W V Type 1 and Type 2 Administrators. MD Comput. 1990; 7(2): 69-70.
Stampfer MJ, Colditz GA, Willett WC, et al, Postmenopausal oestrogen therapy and
cardiovascular disease. N Engl J Med 1991;325:756-62
Tate K E, Gardner R M, Weaver L K, A computerised laboratory alerting reminder system.
MD-Computing. 1990; 7(5): 296-301.
Tierney W M, Hui S L, McDonald C J, Delayed feedback of physician performance versus
immediate reminders to perform preventive care-effects on physician compliance. Medical
Care. 1986; 24: 659-666.
Tierney W M, McDonald C J, Martin D K, et al, Computerised display of past test results:
Effects on outpatient testing. Ann Intern Med. 1987;107: 569-74.
Tierney W M, McDonald C J, Luft F C, Renal disease in hypertensive adults:Effect of race
and Type-II diabetes mellitus. Am J Kid Dis. 1989; 13: 485-493.

Tierney W M, McDonald C J Research using practice databases. Sta Med. 1991; 10: 541-557.
Tierney W M, McDonald C J, McCabe G P Serum potassium testing in diuretic- treated
outpatients: A multivariate approach. Med Decis Making. 1985; 5: 89-104.
Tierney W M, Hannan T J The medical record, clinical care, and the information revolution.
Royal Australasian College of Physicians, Fellowship Affairs. 1992;11:1;17-18
Tierney, WM; Miller, ME; Overhage, JM; McDonald, CJ. Physician inpatient order writing
on microcomputer workstations- effects on resource utilisation. JAMA. 1993; 269: 379-383.
Tierney, WM; Roth, BJ; Psaty, B.; et al. Predictors of myocardial infarction in emergency
room patients. Crit Care Med. 1985; 13: 526-31.
Toward a national health information infrastructure. Report of the work group on
computerisation of patient records. To the secretary of the U.S. Department of Health and
Human Services. April 1993.
Weed L New premises and tools for medical care and medical education. Proceedings
international symposium of medical informatics and education, University of Victoria, British
Columbia, Canada. 1989. p19
Wennberg J; Gittelsohn A. Small variations in health care dleivery. Science. 1973;182: 11028.
Wilson G A, McDonald C J, McCabe G P, The effect of immediate access to a computerised
medical record on physician test ordering: A controlled clinical trial in an emergency room.
Am J Publ Health. 1982; 72: 698-702.