Drug Injormarion Journal, Vol. 17, pp. 205-210, 1983
`Printed in the USA. All rights reserved.
`
`0092-8615/83 $3.00 + .OO
`C 1983 Drug Information Association Inc.
`
`DEVELOPMENT OF A
`COMPUTERIZED DRUG INTERACTION
`DATABASE (MEDICOM””) FOR USE IN A
`PATIENT SPECIFIC ENVIRONMENT
`
`ARTHUR F. SHINN, PHARMD
`Director, Medical Affairs, Professional Drug Systems
`Chairman, Medicom Consulting Group
`St. Louis, .Missouri
`Consultant, Pharmacology, Faith Hospital
`Creve Coeur, Missouri
`
`ROBERT P. SHREWSBURY, PHD
`Assistant Professor of Pharmaceutics, School of Pharmacy
`University of North Carolina, Chapel Hill, North Carolina
`Consultant, Medicom Consulting Group
`
`KENNETH W. ANDERSON, BS, RPH
`Director of Professional Relations
`Medicare-Glaser Corporation
`St. Louis, Missouri
`
`Drug Interactions are a clearly defined problem that we as health professionals must
`deal with on a day-by-day basis. It is by far the area of health care that demands more
`ottention today and tomorrow than was possible in the past. The amount of refer-
`ence sources and text material of drug interactions is growing at such a rate that it
`is almost impossible to recall essential in formation in a reasonable time frame. If the
`practitioner is to continue with an uninterrupted work flow and still maintain the best
`possible service for the patient, an immediate and accurate method is needed in the
`hands of the user. . . a computerized drug interaction database. The community
`health care standards would be ultimately raised to a level never before attainable and
`efficiency would continue with full utilization of professional practice. Combine then,
`drug interaction data with ancillary benefits such as cost containment, third party ac-
`counting, inventory control, and a multitude of other operational functions into a
`computerized database and the end product results in an enhanced, controlled, pro-
`fessional operation.
`
`Key Words: Drug interaction database; Computerized database; MEDICOMsm; Pro-
`fessional Drug Systems
`
`Reprint address: Dr Arthur F. Shinn, Director of Med-
`ical Affairs, I’rofessional Drug Systems, Inc, 2320
`Schuetz Road, St. I.ouis, MO 63146.
`
`IN 1976 Professional Drug Systems, Inc, a
`subsidiary of Medicare-Glaser Corpora-
`a
`tion, began the project Of
`computerized pharmacy system. During
`the conceptual stages of development it was
`
`Downloaded from
`
`205
`dij.sagepub.com
`
` by guest on April 6, 2015
`
`CFAD VI 1020-0001
`
`

`

`206
`
`A . F. Shinn, R . P. Shrewsbury, and K . W . Anderson
`
`assumed that collecting patient prescription
`profile information and not reviewing it for
`certain interactions would be a great waste
`of a pharmacist’s talent and drug informa-
`tion resources. It was therefore determined
`that prescription drug-to-allergy, prescrip-
`tion drug-to-food, prescription drug-to-
`over the counter (OTC) drug, and prescrip-
`tion drug-to-prescription drug interactions
`should be evaluated prior to filling each
`prescription.
`The search was then initiated to identify
`and evaluate the different drug interaction
`systems that were available at that time.
`There were three inherent problems with
`each system that was evaluated:
`
`1.
`
`2.
`
`3.
`
`The systems were based on a philosophy
`that the more information on each inter-
`action the better, ie, quantity, but not
`necessarily quality.
`The databases were reviewed and updat-
`ed by staff which were generally chang-
`ing on a regular basis and the ongoing
`integrity of the system at times was in
`question.
`None of the systems were developed to
`work as an online system to be used in
`an active practice environment in a re-
`tail pharmacy setting.
`
`To overcome these shortcomings, Medi-
`care-Glaser Corporation created a perma-
`nent consulting group and developed an in-
`teraction database to meet the following
`goals:
`
`1.
`2.
`3.
`
`Improve patient care
`Improve professional image
`Supply interaction information on a
`chemical ingredient basis
`4.
`Update the information on a regular
`basis
`Gear the use of the system to the busy
`professional
`
`5 .
`
`The key to the MEDICOM‘” system is
`that the information is set up for use by the
`busy professionals. For example, the health
`professional sees only interactions on cur-
`
`rent prescriptions in the patient’s profile.
`Interaction messages give the specific chem-
`ical ingredients that are interacting, and a
`corresponding significance code and rec-
`ommended action code.
`Table 1 gives the significance codes of
`the system. A code of 1 is for the rare, sel-
`dom occurring interaction, whereas a 9 is
`for an interaction that occurs regularly and
`predictably. The system also includes ac-
`tion codes that range from an A which
`means “Do not dispense- contact prescrib-
`er,” to a Z for those involving nonsignifi-
`cant interactions. Although an action code
`is a recommendation, the practitioner still
`uses professional judgment in the final uti-
`lization of the system’s information. These
`two codes (significance and action) allow
`a health professional to rate each individual
`
`TABLE 1
`Significance Codes
`
`Code
`
`Definition
`
`7
`
`3
`
`5
`
`7
`
`9
`
`These interactions seldom occur
`and are considered not to be sig-
`nificant at this time.
`These interactions have been re-
`ported to occur 20 to 30% of the
`time. Although not frequent, they
`may be significant for a small
`number of patients. Interactions
`classified as a 3 may not have
`adequate documentation and may
`require further study.
`These interactions can occur often
`enough in the population to re-
`gard them as being potentially
`significant. Patient’s physical con-
`dition and other therapy may en-
`hance the significance.
`These interactions have been docu-
`mented in the medical literature,
`can be expected to occur fre-
`quently, and are to be regarded
`as potentially serious. A review of
`the patient’s therapy may be ad-
`visable.
`These interactions occur regularly (al-
`most in every patient), have substan-
`tial documentation, and are to be re-
`garded as being highly significant.
`Change in choice of drug(s) or
`drug(s) dose is highly advisable.
`
`Downloaded from
`
`dij.sagepub.com
`
` by guest on April 6, 2015
`
`CFAD VI 1020-0002
`
`

`

`Development of a Cotnputerized Drug Interaction Database
`
`207
`
`interaction and to react where appropriate
`regardless of the minute by minute primary
`health care activity.
`Along with the significance and action
`codes, the reference source of the interac-
`tion information and the corresponding
`page number in that reference are dis-
`played. There are three reference sources
`used in MEDICOM'": Evaluations of
`Drug Interactions (EDI),' EDI's Supple-
`ment, and the Medicom Drug Interaction
`Manual. This Medicom Drug Interaction
`Manual is designed specifically for the
`MEDICOM'" system and was written and
`is updated regularly by the MEDICOM'"
`Consulting Group. The manual contains the
`monographs of new interactions appearing
`in the primary literature since the publica-
`tion of EDI and the EDI Supplement, or
`more recent information on interactions al-
`ready reported in EDI or its supplement.
`The manual monograph summarizes the
`currently available medical literature in
`readable format and language. Citations of
`the literature sources used in the mono-
`graph preparation are also given in the
`monograph.
`In a practice environment each patient
`becomes part of the MEDICOM'" system
`by filling out a Patient Information Sheet.
`This sheet contains basic demographic in-
`formation (name, gender, birthdate, and so
`on) and relevant medical information need-
`ed to screen prescriptions prior to fill-
`ing such as allergies, disease states, and
`frequently taken over-the-counter medi-
`cations.
`When a new prescription order is en-
`tered into the MEDICOM'" system, each
`chemical item in each drug is identified by
`the system and compared to the chemical
`item(s) in the prescription drugs currently
`active in the patient's profile. If a particu-
`lar chemical combination is identified in
`the interaction database, a message appears
`on the CRT and in hard copy, notifying the
`health professional of the interacting chem-
`icals, the significance and action codes, and
`reference and page number. In addition,
`the interacting prescription numbers are
`
`available for health professionals' review if
`required. Simultaneously, if the patient is
`allergic to the particular chemical(s), the
`system will provide a message of this situ-
`ation.
`In addition to the prescription drug-
`to-prescription drug and prescription
`drug-to-allergy interaction matrices in the
`MEDICOMsm Computerized System, the
`system contains prescription drug-to-food
`and prescription drug-to-OTC drug warn-
`ing information.
`The presence of food can change the ab-
`sorption and/or bioavailability of some
`drugs. If this interaction is possible with the
`chemical(s) in the prescription, a three digit
`prescription drug-to-food warning code ap-
`pears on the CRT display. The three digit
`code simultaneously gives the health pro-
`fessional the manual page number of the
`appropriate monograph which documents
`the interaction, and whether the absorption
`and/or bioavailability of the chemical is in-
`creased, decreased, delayed, or otherwise
`unaltered by the food.
`The prescription drug-to-OTC drug
`warning code directs the health profession-
`al to the OTC section of the manual. This
`section contains the prescription chemical
`matched to the chemicals that are known
`to interact with over-the-counter products.
`Also available in the listing are the signifi-
`cance and action codes as well as the refer-
`ence and page number. The health profes-
`sional can then caution the patient on
`which over-the-counter products should be
`avoided while taking this particular pre-
`scription medication.
`The MEDICOM'" System was moni-
`tored for one calendar year (April 30, 1981
`through May 1, 1982) to determine the
`number of prescription drug-to-prescrip-
`tion drug interactions detected by the data-
`base. Four pharmacies in the Medicare-
`Glaser Corporation that have been on the
`system for a year were selected. These
`stores are in different geographical areas
`with very different clientele.
`Table 2 shows the total number of pa-
`tients, the number of original prescriptions
`
`Downloaded from
`
`dij.sagepub.com
`
` by guest on April 6, 2015
`
`CFAD VI 1020-0003
`
`

`

`208
`
`A . F. Shinn, R . P . Shrewsbury, and K . W. Anderson
`
`TABLE 2
`MEDICOMU" Drug Interaction Study From 4-30-81 Thru 5-1-82
`- _..
`- _.
`Sample
`Number of
`Store
`Patients
`
`#1
`#2
`#3
`#4
`TOTAL
`
`4,978
`10,270
`4,071
`4,258
`23,577
`
`Number of
`Prescriptions
`__
`21,786
`40,462
`14,235
`19,390
`95,873
`
`
`
`Number of Drug
`Interactions
`_
`_
`. _
`3,759
`6,632
`2,124
`4,065
`16,580
`
`- -
`
`-
`
`entered into the system, and the number of
`prescription drug-to-prescription drug in-
`teractions recorded during the study peri-
`od. The number of original prescriptions
`included those prescriptions entered into
`the system but not filled due to the inter-
`action(s) detected.
`The distribution of the prescription drug
`interactions by their significance code is
`shown in Table 3. In these four store sam-
`ples, 33% of the interactions detected were
`highly significant and were usually ex-
`pected to occur in patients (codes 7 and 9).
`Over 50% of the interactions detected were
`either significant or highly significant (codes
`5, 7, and 9).
`When the number of interactions detect-
`ed was related to the number of patients
`having prescriptions entered into the sys-
`tem (see Table 4), 23.2% of all patients had
`an interaction of significance code 7 or 9
`regardless of how many prescriptions they
`had entered. Considering the number of in-
`teractions detected versus the number of
`original prescriptions entered in the system,
`
`17.3% of all prescriptions were expected to
`have some type of interaction, and 5.7% of
`those interactions had a significance code
`of 7 or 9.
`Table 5 shows the number of original
`prescriptions entered in the system for each
`patient. 37.2% of the patients had only one
`prescription entered, which accounted for
`8,759 of the 95,873 prescriptions entered
`during the study. These prescriptions would
`not produce a prescription drug interaction
`entry because only one prescription is in-
`volved. Therefore, only 14,818 patients
`would have the situation of a prescription
`drug interaction because they had more
`than one prescription entered into the sys-
`tem. Also, there would be only 87,114 pre-
`scriptions entered for these 14,818 patients
`and these produced 16,580 interactions.
`Table 6 shows the significance code distri-
`bution of the interactions based on these
`14,818 patients and their prescriptions. The
`results suggest that for every patient hav-
`ing more than one prescription entered in-
`to the system, there were 1.12 prescription
`
`TABLE 3
`Interaction Percent Distribution by Significance Code
`Significance Code (See Table 1)
`
`Sample Store
`
`#1
`#2
`#3
`#4
`Study Average
`
`01
`
`03
`
`07
`05
`- _.
`18.8 30.7
`14.3 29.2
`19.3 25.6
`17.3 31.8
`30.9 19.8 26.0
`16.1
`13.5 36.9 18.9 25.1
`15.5 32.3 19.2 26.7
`
`09
`
`7.0
`6.0
`7.2
`5.6
`6.3
`
`Downloaded from
`
`dij.sagepub.com
`
` by guest on April 6, 2015
`
`CFAD VI 1020-0004
`
`

`

`TABLE 4
`interactions as a Percent of Patients and Prescriptions
`by Significance Distribution
`
`Significance Code (See Table 1)
`
`Total
`
`Number of
`interactions Total Oh 1 (010) 3 (010) 5 (Oh) 7 (Vo)
`
`9 (010)
`
`Number of
`Patients
`Number of
`Prescriptions 95,873
`
`23,577
`
`16,580
`
`16,580
`
`70.3
`
`17.3
`
`10.9
`
`22.7
`
`13.5
`
`18.8
`
`2.7
`
`5.6
`
`3.3
`
`4.6
`
`4.4
`
`1.1
`
`TABLE 5
`Distribution of Number of Prescriptions Per Patient
`
`Number of Prescriptions per Patient
`
`1 Rx
`
`2 R x
`
`3 R x
`
`4 Rx 5 Rxormore
`
`Number of Patients 8,759 5,300 2,896 1,937
`Percent of Patients 37.2% 22.5010 12.3% 8.1%
`
`4,685
`19.9Oh
`
`TABLE 6
`interactions as a Percent of Patients and Prescriptions
`by Significance Distribution
`
`Significance Code (See Table 1)
`
`Total
`
`Number of
`Interactions Total Oh 1 (010) 3 (010)
`
`5 (010)
`
`7 (Oh) 9 (010)
`
`Number of
`Patients
`Number of
`Prescriptions 87,114
`
`14,818
`
`16,580
`
`111.9
`
`17.3
`
`36.1
`
`21.5
`
`29.9
`
`16,580
`
`19.0
`
`2.9
`
`6.1
`
`3.6
`
`5.1
`
`7.0
`
`1.2
`
`209
`
`Downloaded from
`
`dij.sagepub.com
`
` by guest on April 6, 2015
`
`CFAD VI 1020-0005
`
`

`

`210
`
`A . F. Shinn, R. P. Shrewsbury, and K . W. Anderson
`
`drug-prescription drug interactions detect-
`ed. Stated another way, patients with more
`than one prescription entered in the system
`will have an average of 19% of those pre-
`scriptions interacting in some manner.
`In 1981 1.4 billion prescriptions were
`filled in the United States.2 If the experi-
`ence of the four pharmacies included in the
`study were typical of the nation, then 17.3%
`of the prescriptions filled would have some
`interaction regardless of how many pre-
`scriptions each patient had filled. There-
`fore, 242 million interactions would be de-
`tected by a system such as MEDICOMsm,
`and 79.8 million of those interactions would
`have a significance code of 7 or 9. Thus, 1
`out of every 6 prescriptions filled in 1981
`had an interaction of some manner, and 1
`out of every 18 prescriptions filled had an
`interaction with a significance code of 7
`or 9.
`In the four pharmacies in this study,
`62.8% of the patients had more than one
`
`prescription entered into the system (see
`Table 5). Hence, 62.8% of the prescrip-
`tions filled were for patients with more
`than one prescription and as a result ap-
`proximately 879 million prescriptions are
`assumed filled for patients with more than
`one prescription. As assumed in the previ-
`ous paragraph, prescriptions filled in 1981
`would have accounted for approximately
`242 million drug interactions. Therefore,
`patients with more than one prescription
`filled in 1981 will have an average of 27.5%
`of their prescriptions interacting in some
`manner, with 9% of those interactions hav-
`ing a significance code of 7 or 9.
`
`REFERENCES
`
`I . Evaluations of Drug Interactions, ed 2. American
`Pharmaceutical Association, Washington, DC,
`1976.
`2. Kushner D: Self-pay prescriptions decline for first
`time, survey finds. American Druggist, May 1982;
`185: 11-14.
`
`Downloaded from
`
`dij.sagepub.com
`
` by guest on April 6, 2015
`
`CFAD VI 1020-0006
`
`