MedicalTranscription.com
Electronic Medical Records Primer
Content provided courtesy of
Kirk G Voelker MD
Please see below for
additional information
Electronic Medical Records Primer
Contents:
Introduction
What is a medical record?
A medical record is a confidential
record that is kept for each patient by a healthcare professional or
organization. It contains the patient's personal details (such as name,
address, date of birth), a summary of the patient's medical history, and
documentation of each event, including symptoms, diagnosis, treatment and
outcome. Relevant documents and correspondence are also included.
Traditionally, each healthcare provider involved in a patient's care has
kept an independent record, usually paper based. The main purpose of the
medical record is to provide a summary of a person's contact with a
healthcare provider and treatment provided to ensure appropriate
healthcare.
Information from medical records also
provides the essential data for monitoring patient care, clinical audits
and assessing patterns of care and service delivery. In the current
environment the medical record also forms the first link in the
information chain producing the depersonalized aggregated coded data for
statistical purposes.
As every health professional, coder,
manager and patient knows, considerable effort is invested in writing,
filing, sorting, searching, retrieving, issuing and recovering the medical
record, in whole or in part. There is no doubt that the ready availability
of well organized, legible, accurate and comprehensive clinical notes can
play a very significant role in the clinical decision making process and
assisting in the provision of quality healthcare.
The medical record should:
 |
 enable
health professionals to review previous care events, to reach timely and
appropriate clinical decisions, and to develop treatment plans that
minimize the risks and maximize the potential benefits to the patient
|
|
provide
an archival and legally acceptable record of the steps that were taken -
when, why and by whom - in the care of an individual
|
|
enable
staff to audit the care provided to an individual
|
|
provide
material for researchers studying the etiology, natural history and
cost-effective approaches to treatment of specific conditions
|
|
act
as a source of information which will enable various administrative
functions of the healthcare service unit (such as contract management or
coded statistical returns) to be carried out automatically as a
by-product of the clinical data collected
|
|
be
stored in such a way as to ensure that the data are secure from loss,
alteration or damage |
|
be
subject to access controls that ensure patient privacy is adequately
protected, and that the risk of disclosure to unauthorized persons is
minimized.
|
Given the changes in technology
particularly the move to computerized information storage and increasing
consumer or patient involvement in healthcare, one issue that must be
addressed is whether the existing paper-based medical record remains the
most cost-effective way of achieving these goals.
|
Shortcomings of the
paper record
A distinguished American physician has
commented:
"The (paper) medical record is an
abomination ... it is a disgrace to the profession that created it. More
often than not the chart is thick, tattered, disorganized and illegible;
progress notes, consultant's notes, radiology reports and nurses notes
are all co-mingled in accession sequence. The charts confuse rather than
enlighten; they provide a forbidding challenge to anyone who tries to
understand what is happening to the patient."
Bleich, H., MD, Computing Vol 10 no 2, p70, 1993.
Another view is:
"...medical records, which have long
been faulty, contain more distorted, deleted and misleading information
than ever before."
Burnum, J.F. The misinformation era: the fall of the medical record.
Annals of Internal Medicine 110:482-4, 1989.
Certainly the paper record has been seen
by many to have serious shortcomings in dimensions other than the
historical record of care. For example:
Content: finding and re-using data
items
It is difficult to find specific items
of information in it (for example, an entry, a report). Different users
need different kinds and levels of information. Important information may
be hidden amongst the clutter of trivia. Key pointers may be missed.
Information may simply not have been collected or recorded, or may have
been misplaced. Similarly the process of coding for contracts and
statistical returns may be seriously hampered by the difficulty of finding
key items of information. From the point of view of re-use of information
for research or administration, it has been said that once data are
committed to the paper record, they are entombed there for ever. Certainly
the costs of abstracting data from records can be very high.
Record fragmentation
The growing need to share the care of
patients between healthcare providers (for example, between GP and
consultant, between one clinic and another) is also often poorly served by
the paper record. It can only be in one place at a time and logistical
issues make it difficult to move it around as fast as is needed. In
practice every healthcare unit has a separate record for each individual,
thereby creating a serious problem of record fragmentation and
dis-integrity. This can lead to potentially serious problems of continuity
of care for the patient. It also threatens the freedom of patients to
choose where they go for care, as well as their right to equity in access
to appropriate care.
Moving records
Electronic exchange of patient
information is rapidly developing, because of the potential to save time
and money. But in order to take full advantage of this opportunity it is
highly desirable that the data are stored in electronic format - otherwise
the record must be rekeyed, with all the attendant issues of extra work
and transcription errors.
Decision support systems
Interest is fast growing in the use of
decision support systems. These assume a very wide range of forms, but
can, for example, provide expert critiques of clinical decisions, indicate
the most cost-effective options for further work-up, investigation and
management, or provide context-specific knowledge from the biomedical
literature. Decision Support Systems also have a major role to play in
quality assurance as well as in cost containment.
|
Looking forward:
knowledge-based tools
The sheer size of the medical knowledge
base poses a forbidding problem for the would-be healthcare practitioner.
Opinions vary, but it is believed that there are upwards of 5,000,000
medical facts. New research published in more than 30,000 biomedical
journals adds to this number, whilst at the same time discrediting,
altering or extending some of the other facts. The average student can
master perhaps 5-10 facts an hour under good conditions, but rapidly loses
knowledge that is not reinforced by regular use. The calculation is
simple: there is in excess of 200 years of full time study just to master
medical knowledge as it stands today.
The safe and cost-effective practice of
medicine is becoming increasingly complex, and relies more and more on
knowledge of the results of recent research into causes, manifestations,
diagnosis and effective treatment of illness. To quote another
distinguished American physician:
"I once believed it was possible to
recall and process all the necessary variables at the bedside or in the
clinic at the time of clinical action after one has mastered a core of
medical knowledge. I now believe that we need electronic extensions of
the human memory and analytic capacity just as we need X-rays to extend
the human eye. Physicians should rely on such tools just as a traveler
relies on maps instead of memorizing trips in geography courses.
Faculties and researchers should keep the tools up to date."
Weed, L.L. Perspectives over 40 years. Proceedings of the ACM Conference
on the History of Medical Informatics, p 105. New York: ACM 1987.
Knowledge-based decision support tools
are increasingly being used and are likely to prove invaluable as a means
of ensuring and assuring best quality care and practices for all patients.
There are already in excess of 1000 medical 'expert systems' in use around
the world. However, the only efficient way of using them is to hold the
medical records in an electronic form with which these tools can interact
directly.
|
Investment in
medical records
Recent studies by the United Kingdom
Audit Commission (1994) have indicated that the cost of medical records
units runs at between 2% and 6% of turnover: this figure takes no account
of clinical or analytical staff time spent on recording, searching,
processing and analyzing the record contents, since hard data on this
issue are difficult to obtain. Some staff claim to be spending as much as
75% of their time chasing items of information relating to patients and
written on pieces of paper, and there can be very few healthcare providers
who spend less than 20% of their time reading, writing, sorting and
searching through the notes.
There is no dispute that medical records
are a valuable resource but they also involve expenditure of large amounts
of time and money: often that expenditure is not explicitly recognized or
quantified, but is lost amongst the general overheads. Questions that are
asked include: Could that money be better spent, or even reduced, by a
switch to electronic records? Could the professional time involved be
reduced? Could the medical records become more functional and useful?
|
What are electronic
medical records?
It is possible to store the entire
medical record, or any part of it, on computer. To make such a change
involves significant investment in both equipment and staff, but the
benefits can far outweigh the costs, as well as setting a course for the
future.
What is important in looking at the
opportunities presented by electronic medical records is to recognize that
we are not comparing like with like. Electronic records greatly extend the
concept of the medical record, and enable many functions that are
otherwise quite impossible.
What are the benefits?
Availability, transfer and retrieval
Electronic files can be readily
accessed from anywhere, local or remote, across a communications link or
network. Data that are stored in electronic formats can be retrieved
electronically: literally billions of records can be sifted through in
seconds if the database has been appropriately designed and indexed.
More than one user at a time can have access to them, and all service
providers can share the same records.
Linkage
Records made by multiple providers in
different locations and units can be linked and shared to create a
single record for the individual. The problem of record fragmentation
can be resolved, and patient care can be genuinely shared between
providers. Further all the graphic data (for example, images), incoming
letters (for example, referrals) and auditory data (for example, heart
sounds, spoken notes) relating to a patient can be linked to their
electronic record file using multimedia techniques.
Storage
Electronic storage of data is cheap
and very compact. A single compact disc (CD) can store in the region of
600 Mbytes, equivalent to some 100,000 pages of text or about 150 large
textbooks which would need about 10 meters of shelf space. The CD is
almost indestructible.
Data views
One of the special benefits of
computerized records is their ability to readily display different views
- for example, all current medications, or problems; the last ten full
blood counts in graphic display; test results for a specified admission
or date range. The data in the record are no longer static and
accessible only in the order and format determined by the writer, but
can be dynamically displayed in any way that suits the needs of
different viewers.
Abstraction, reporting
Once in electronic format, records can
be reported upon automatically. Patients' treatments can be assigned to
DRGs, statistical reports can be sent to the national collections,
notifications (for example, of births and deaths) can be sent to the
Registrar-General and so on. Automatic audit reports can be prepared,
for example of caseloads, services provided, lengths of stay, costs of
care and so on. Data can quickly be gathered for research studies and up
to the minute reports generated. All the data required for
administration and contract management can be derived automatically from
the medical records.
Data quality and standards
Data can be checked as they are
entered to ensure adequacy (for example, for casemix and contract
management) and accuracy by querying entries that are unlikely (for
example, heart rate 200-220) or rejecting those that are impossible (for
example, plasma sodium outside the range 120-160 meq/L). Results and
reports can be entered directly from other systems, eliminating the
possibility of misfiling and of transcription errors.
Decision support
Direct links to knowledge-based tools
can be built: the present development of the Arden syntax and a library
of Medical Decision Logic Modules will make it possible for any system
to incorporate intelligent alerting flags to users warning them of
possible errors, and advising on the best way forward.
In very practical terms there are major
advantages to the health professional/clinician in being able, at the
press of a button, to automatically retrieve and repeat prescriptions,
fill in forms (for example, for tests and investigations) accurately and
automatically, complete and send discharge summaries and so on. There are
also significant advantages to having access to the medical record,
whether generated locally or in some remote care centre. The advantages to
administrators are clear in that quality data for reporting, workloads,
costings and audit are readily accessible. The benefit to the patient is
that of continuity and integrity of care wherever the patient may be. The
benefit to the community is of delivery of best quality care in the most
economic ways possible.
In 1991 the United States Institute of
Medicine published 'The computer-based patient record - an essential
technology for healthcare' (Eds Dick, R.S. and Steen, E.B. Washington:
National Academy Press). This concludes that there needs to be immediate
adoption of the concept and substantial investment in its development for
use across the nation by the year 2000, for many of the reasons outlined
above.
|
Obstacles to
adoption
While the benefits are clear, there are
some obstacles. These, however, relate mainly to practices and people and
much less to technology. All the technology that is required to create
electronic medical records systems already exists, and, indeed, electronic
records are already in everyday use in many parts of the world. In broad
terms the main requirements are that there should be:
confidence in computers amongst their
users, and especially in the availability, privacy and security of data
stored on computer active clinical use made of the electronic medical
records; adoption of a positive attitude towards computers in the
workplace. This may require careful strategic management of change, as
well as champions for the new technology recognition and acceptance by
those entering data that the usefulness of records extends beyond the
needs of care of the patient: that there are many legitimate re-uses of
clinical information which are in the best interests of the community as
well as of various other parts of the organization (for example, billing,
research, statistics) adequate skills and proficiency in the use of the
computer application; users should become knowledgeable consumers of this
technology (like a motorist) without needing to be experienced in its
maintenance (like a mechanic). A level of basic knowledge and
understanding about computer systems will be invaluable, especially for
those who may have to manage their own installation, but advanced keyboard
skills are generally quite unnecessary.
|
What is in the
record?
The record comprises a mix of written
text, codes and images. There is no reason why, in the future, it should
not contain audio memos and notes as well. Much of the material in the
record has a relatively short useful life - who is interested in whether
the patient was pale 3 years ago? Other material is much more durable -
for example details of admissions, diagnoses, major procedures.
The material in the medical record can
be separated into three quite different components:
|
Material
which comes from elsewhere (for example, a referral letter, or a set of
biochemical measurements), and which is supplied in an 'as is' form.
This may include incoming letters and reports, pictorial (ECG traces,
sketches, etc) and image material if provided in hard copy form
only. This is generically referred to in this paper as 'pictorial'
material. |
|
Material
generated locally, mostly as a consequence of a care encounter. This is
of two forms:
-
'Free
text', that is an unstructured description of findings,
-
conclusions
and plans using any words chosen by the writer
-
Material
that is structured in one or more of the following ways:
- organized around a data entry template (for example, check boxes)
- using a controlled vocabulary (for example, picked from a list of
acceptable terms)
- using a system of classification and coding (for example, Read codes
or ICD).
|
The reason for making this distinction
is that each has to be treated in a quite different way, and each imposes
different constraints on the capacity for and technology required to
transfer, sort, search and manipulate the record.
Pictorial material
The issue with pictorial material as a
category is that it can be incorporated into the record only as an
electronic image. The letter has to be 'scanned', in a manner similar to a
photocopier or fax machine, and an electronic 'bitmap' of the arrangement
of marks on the page stored as a file. It can be linked to a patient
record as an 'object' associated with that record, but it cannot readily
be searched, for example when preparing reports or research studies.
Flatbed scanners are available for Document Image Processing (DIP) which
converts these materials into electronic image files. Scanners have
differing resolutions, typically 200-300 dpi (dots per inch), depending on
the resolution required.
However, if the material is transferred
in an electronic format, it may be possible to incorporate it into the
record as free text, or even as structured material. For example a
laboratory result that is transferred electronically can be incorporated
into the patient record as an entry that can be searched and analyzed, and
even displayed graphically as a time sequence (for example, of changes in
platelet counts).
'Free text'
Natural language in the record, such as
a sentence of this paper, can be searched for the occurrence of 'strings'
of characters, that is, for an exact match with a specified sequence of
letters or numbers.
One of the main purposes in storing much
of the data in the record is in order to report upon it. Being able to
find it with a search is therefore vitally important. In this regard
string searches have certain shortcomings:
|
They
are slow and cumbersome, the more so the larger the database.
|
|
If
the word has been mis-spelled (on average 1 keystroke in 20 is
erroneous), the search fails to find the entry.
|
|
If
a different word meaning the same (synonym) was entered, the search will
not find it (for example, AMI or heart attack in place of myocardial
infarct). Synonyms abound in medicine, including eponyms (use of proper
names such as 'Pagets disease'), and few people are consistent in their
use of terminology. |
|
If
an abbreviation has been used, the search will not detect it.
|
|
If
the entry has been made in the wrong field (see structured text below)
of the database (for example, the diagnosis has been entered where the
system expects the complaint to be stored), the search will fail to
detect it. |
|
If
a search is required for a group of entities (for example, all renal
diseases, all medications) it is a very long and difficult (probably
near impossible) job to string search for each of the possible terms
that might have been used to describe any of the hundreds of renal
conditions or medications.
|
In summary, free text can be useful in
electronic records where the data will simply be retrieved for display or
printing. Whilst some limited manipulation (searching, analysis) is
possible, it is difficult and unreliable and should be avoided.
Structured text - the options
Structuring the text makes the problem
of searching much more effective and efficient. Structure can be
introduced by:
| providing a framework within which
the data are to be entered (for example, checkboxes);
|
| limiting the entries to a set of
'preferred' terms ('dictionary' linkage) and rejecting any entries not
in that dictionary; |
|
displaying
the alternatives from which to select for each box (a 'picking' list);
or |
|
requiring
that entries are classified (and coded).
|
Option 1 means that the computer knows
where to search for specific data items; however, within each field it
will still have to string search. Option 2 eliminates some of the problems
of string searches, in that entries are validated against the dictionary
before being accepted. This may be frustrating for a user who does not
know the terms in the dictionary. It also still requires the user to type
the entry and to spell it correctly. Option 3 further validates the
entries ensuring that only an appropriate term is entered (for example,
only diagnoses in the diagnosis box). It also overcomes the need to learn
the dictionary since only acceptable options are displayed, and the need
to type, since items may be picked with the mouse.
Option 4 optimizes the record for
computer manipulation: codes are much easier to search and sort than free
text. Depending upon the implementation, it may impose some entry time
overhead, but this can be minimized by using picking lists (see 3 above)
which display the text, whilst entering the relevant code in the database.
Some entities (locations/wards, local administrative measures,
sub-specialty specifics, etc) may not have codes in formal classifications
(such as Read Codes or International Classification of Diseases (ICD)):
these will need a local code set developed if they are to be used for
searching and reporting.
Data items which are most likely to be
re-used are:
|
problem/complaint
and/or diagnosis |
|
investigations
|
|
medications
|
|
services
|
|
professionals
attending |
|
dates,
times and locations.
|
Data that are not required for re-use
can be held as free text (see above).
In summary, all data that will be used
for analysis, or as a key for searching and sorting the record should be
held in a structured, and preferably in a coded form.
|
| |
Electronic data
exchange
It is possible to communicate
information of almost any type electronically using computers. Some
communications involve very large amounts of data (for example, a picture
or X-ray) whilst others are relatively small (for example, a memo). In
every case the message to be sent is translated into a sequence of 'bits'
or 'digits', sent across an appropriate transmission medium, such as a
telephone line, optical fiber or radio link, and returned at the receiving
end into its original form.
Once medical record information has been
stored in an electronic format, it can readily and automatically be
exchanged in this way. There are a variety of technical requirements, but
these are relatively easily accommodated. The communications protocol that
has been adopted for these interactive messages is an industry 'standard'
called HL7, and all messages can be sent over the New Zealand Health
Information Service network operated by third-party Value Added Network
(VAN) services.
All messaging environments require the
details of message structures and syntax to be spelled out in precise
detail in order to ensure that there can be no misunderstandings between
sender and receiver. The message for advising a purchaser of a healthcare
event and billing for it, for example, will be very different to one
reporting the results of a pathology test to a provider. Any given message
can be generated automatically from an electronic medical record, as long
as the required data elements are stored in the record. This is done by a
software routine which is designed to find the required elements from the
medical records and place them in the correct place in the message. In
just the same way an incoming message can be analyzed and the data
elements it contains automatically dropped into the correct slots in the
medical records system.
|
General practice
and hospital records
The division of medical or health
records for an individual based on the identity, specialty and/or location
of the provider is a serious and growing problem. The subject of the
record is the individual, and that should be the focus and organizer. Only
when the community and hospital care records for an individual can be
brought together will much of the essential purpose of keeping these
records for integrity and continuity of patient care be achieved.
Whilst the structure of two records
systems and their layout may appear to be quite dissimilar, there are
significant parts of the content, however arranged, that are closely
aligned. For example, both records will contain identifying and personal
data (name, address, date of birth, etc), clinical problems and diagnoses,
current medications, test and investigation results and so on. All these
data can be successfully aligned and exchanged using electronic messages.
This could replace the existing referral and report letters which are the
present means of communicating these data.
|
Data entry
For many potential users, the issue of
data entry is a source of anxiety: will everyone involved in patient care
need to have advanced keyboard skills? The answer is definitively "no",
although possession of basic keyboard familiarity, as well as resolution
of technophobia may be a significant advantage in some circumstances.
If the medical record contains large
amounts of 'free text', someone has to write it, and this does indeed
require typing skills. However, it may be possible to arrange for a
secretary to enter the data from notes (written or audio) made by the
provider.
Where the data in the record are
structured, data entry can be greatly facilitated by the use of checkboxes
or picking lists: these require minimal keyboard skills. Placed in the
context of a data collection protocol, the entire data set associated with
a specific care encounter (for example, an antenatal care visit, an
assessment for cardiac surgery) can be collected without any conventional
typing at all.
There are various other ways of entering
data, for example:
|
optical
mark recognition (OMR), where a mark on a standard form is sensed by a
reader and converted into a data item in the record
|
|
bar
coding, where the data are stored as a sequence of bars of variable
width and separation and can be detected with a light pen or scanner.
|
Both of these have important roles to
play in the development of full electronic medical records systems that
are easy to use.
There is often talk of the potential of
'voice recognition' systems. Whilst great strides in voice recognition
have been made in recent years, there would seem to be little prospect in
the immediate future for cost-effective systems suited to the needs of a
hospital where there may be hundreds of staff needing to write material,
all using different accents and dialects, often in relatively high ambient
noise environments, and where there is a significant need for parsing and
interpretation of the data entered. The advent of multimedia systems does
make it possible to store spoken notes: these have all the same
restrictions on them as pictorial material (see above), making them
unsuited to searching, sorting, analysis or reporting.
|
Summary
It could be argued that there is a
strong business case for the development and implementation of
computer-based medical records. All the necessary technology for
implementation of full electronic medical records exists. Where electronic
records are kept as an integral part of the care planning and delivery
process, their data quality is normally high, and almost all the
administrative requirements for data can be provided as a by-product of
these records. The added benefit is that providers can generate summaries
of encounters and use them for internal quality assurance audit as well as
for review and personal study: summaries of encounters (log books) are
increasingly required by junior doctors as part of their professional
advancement. Data for contract management and for advice to referring
practitioners (for example, discharge letters) can also be automatically
generated at the time of discharge. Without computerized records, many of
the potential benefits of electronic communications will not be able to be
realized.
For the electronic medical record or
computerized patient record to be accepted there are many issues to be
worked through. The workplace needs to be carefully prepared for their
introduction if the change is to be successful. Acceptance by providers of
the wider purposes of medical records (beyond patient care) is essential.
Many software applications exist, but few of them are entirely
satisfactory. The key questions to resolve are: who will enter
which data, when, where, using what technology,
and for what purposes.
The structure of the medical record is
another issue since there is as yet no international standard for the
electronic medical record. To be successful, core data elements should be
entered within a structured environment (for example, a data collection
protocol), using checkboxes or picking lists. All data items which are to
be re-used for searching, sorting, analysis or reporting should be
classified and coded using either accepted external coding systems (for
example, Read, ICD) or internally maintained coding systems (for
personnel, locations, administrative issues, etc). The use of bar codes
and optical mark readers for data entry may be invaluable in specific
circumstances. All record entries are comprised of events: an event
in this context normally comprises an element (for example, blood
pressure, or a diagnostic decision) a value (for example, 130/75 mm Hg, or
diagnosis of acute appendicitis), a time/date stamp, and an attribution
(for example, Dr Lee Jones made this entry). It may also include an entry
place/terminal identifier. These events may be grouped together into
encounters and episodes of care according to the paradigm(s) prevailing at
that time or imposed variously by different purchaser organizations.
Data that are less likely to be re-used
may be collected in less structured ways. However, the development of
records systems based entirely on free text and pictorial scanned material
provides little added utility over paper records, except that they may be
more easily located, accessed and kept together.
More importantly the legal and social
frameworks for the acceptance, introduction and implementation of
electronic patient records need to be developed, debated and established.
This is an urgent task.
|
Content provided courtesy of Kirk Voelker, M.D.
For more information about Electronic Medical Records please visit his site:
Electronic Medical Records
|
