International Journal of Medical Informatics 60 (2000) 21 – 28

www.elsevier.com/locate/ijmedinf

The Mosoriot medical record system: design and initial

implementation of an outpatient electronic record system in
rural Kenya
Terry J. Hannana, Joseph K. Rotichb,c,e, Wilson W. Oderob,c,e, Diana Menyae,
Fabian Esamaie, Robert M. Einterzc, John Sidlec,e, Joy Sidlec, Faye Smithb,
William M. Tierneyb,c,d,*
a
St. George Pri6ate Hospital, Kogarah, NSW, Australia
b
Regenstrief Institute for Health Care, Indianapolis, IN, USA
c
Department of Medicine, Indiana Uni6ersity School of Medicine, Indianapolis, IN, USA
d
Roudebush VA Medical Center (11H), 1481 West Tenth Street, Indianapolis, IN 46202, USA
e
Moi Uni6ersity Faculty for the Health Sciences, Eldoret, Kenya
Received 16 March 2000; received in revised form 10 May 2000; accepted 15 May 2000

Abstract

Mosoriot Health Center is a rural primary care facility situated on the outskirts of Eldoret, Kenya in sub-Saharan
Africa. The region is characterised by widespread poverty and a very poor technology infrastructure. Many houses
do not have electricity, telephones or tap water. The health center does have electricity and tap water. In a
collaborative project between Indiana University and the Moi University Faculty of Health Sciences (MUFHS), we
designed a core electronic medical record system within the Mosoriot Health Center, with the intention of improving
the quality of health data collection and, subsequently, patient care. The electronic medical record system will also
be used to link clinical data from the health center to information collected from the public health surveys performed
by medical students participating in the public health research programs of Moi University. This paper describes the
processes involved in the development of the computer-based Mosoriot medical record system (MMRS) up to the
point of implementation. It particularly focuses on the decisions and trade-offs that must be made when introducing
this technology into an established health care system in a developing country. © 2000 Elsevier Science Ireland Ltd.
All rights reserved.

Keywords: Computer information systems; International health; Public health

* Corresponding author. Tel.: + 1-317-5540000, ext. 5057; fax: +1-419-7937256.

E-mail address: wtierney@iupui.edu (W.M. Tierney).

1386-5056/00/$ - see front matter © 2000 Elsevier Science Ireland Ltd. All rights reserved.
PII: S1386-5056(00)00068-X
22 T.J. Hannan et al. / International Journal of Medical Informatics 60 (2000) 21–28
1. Introduction
The Indiana University and the Moi Uni-
versity Faculty of Health Sciences (IU-
MUFHS) collaboration began in 1989 with
the purpose of improving the public health of
Kenyans by enhancing medical education
and public health research, the latter through
a program called community-based education
and service (COBES) [1,2]. The ultimate goal
of this collaboration is to influence the devel-
opment of health care leadership in Kenya
and the US. Critical elements of this collabo-
ration include the exchange of ideas and
manpower, not the transfer of money and
technology. All programs introduced as part
of this collaboration must become self-sus-
taining by the Kenyan community.
2. Description of the Mosoriot Health Center
The health center is situated in a rural area
: 25 km (15 miles) from Eldoret. It provides
free medical support for antenatal care, chil-
dren B 5 years and family planning. Patients
are charged small fees (15 Kenyan shillings
each = : US$0.20) for visits to adult
medicine, child care for children \ 5 years of
age and selected services, such as very basic
X-ray and clinical laboratory facilities. Pa-
tients are also charged 15 shillings for each
drug dispensed from the Mosoriot pharmacy
that contains moderate quantities of a small
number of drugs, mostly anti-infectives. The
pharmacy fee must be paid prior to receiving
the treatment. Patients unable to pay are
referred to an outside chemist (pharmacist)
where they may or may not receive the ser-
vice. The only central record of treatments
given is kept in the financial office. Therefore,
no records of treatments are kept for patients
who cannot pay for treatments and patients
who get the treatments free (i.e. pregnant
women, children B 5 years of age and pa-
tients visiting the family planning clinic).
There is no other record of the costs of care
or resource utilization.
The health center is the sole health care
provider for a surrounding population of 30–
40 000 people; it provides care during :
40 000 clinic visits per year. Most of the
population live in small villages with mud-
walled houses and thatched roofs. There is
generally no running water, electricity or
telephones.
On arrival at the health center, each pa-
tient is currently registered by recording his
or her name in a registration book, along
with a visit number, which is the sequential
number of that visit for that year. This num-
ber is reset to one at the beginning of each
year. There is no unique patient identifier
comparable to the social security number in
the US. After registration, patients are
triaged by the medical records clerk who
directs them to the appropriate clinic.
In each clinic within the health center, the
patient is re-registered in a separate book,
again recording the name, visit number, com-
plaint and diagnosis. The same information is
also recorded in a small booklet that each
patient (adult and child) must purchase and
replace annually. This book is the main ‘per-
manent’ longitudinal medical record, the
source where the health care providers seek
information about prior visits. Inherent in
this process of care is the constant re-entering
of patient record numbers, complaints and
diagnoses, much of which is illegible and with
significant transcription errors. Currently the
medical record department stores the records
representing 10 years of patient visits that fit
into an area equivalent to two or three stan-
dard suitcases.
The Mosoriot Health Center is predomi-
nantly an outpatient facility. More critically
ill patients are either admitted to a small,
 T.J. Hannan et al. / International Journal of Medical Informatics 60 (2000) 21–28
20-bed inpatient unit at the Mosoriot Health
Center or referred to Eldoret Teaching and
Referral Hospital or another district hospital.
The Mosoriot Health Center is required to
produce regular reports to the Kenyan Min-
istry of Health on the health center’s activi-
ties. These monthly, quarterly and yearly
reports contain counts of selected reportable
conditions, such as malaria, seen by the vari-
ous clinics. Again, because there are no
unique patient identifiers, these are reports of
visits and not individual patients. Counts of
individual patients cannot be obtained from
these data.
3. Designing the Mosoriot medical record
system (MMRS)
In February and October 1998, and again
in January of 2000, members of the project
team from the US (RME, JS, WMT), Aus-
tralia (TJH) and Kenya (JKR, WWO, DM)
visited the Mosoriot Health Center. They met
with the Health Officer, the matron (head
nurse), director of medical records, nurses
and staff in all offices and clinics. They di-
rectly reviewed the administrative and clinical
activities of all aspects of the health center
along with the health center’s reporting re-
quirements. These visits helped define the
clinical and administrative core data require-
ments for the health center and provided
guidelines as to how the basic record model
should look. They also provided the needed
guidance to develop the data dictionary. We
desired to be as unobtrusive as possible in
order to force the minimum number of
changes on the flow of patients and the tasks
of the Mosoriot staff. The project team there-
fore decided to base the entire MMRS in a
single computer located in the registration
office. Because each clinic records data for
each patient and the patients move from the
23
registration office to the clinics, then back to
the registration office, and finally to the
financial office, the model that seemed to fit
the best was to develop an ‘encounter form’
that the patient will be given at the time of
registration and then carry to each care site
(clinic, laboratory, X-ray etc.). At the end of
the visit, the patient using the new encounter
form will return to the registration office
where selected data from the form will be
entered into the MMRS. The encounter form
will then be given to the patient in lieu of
recording information in his or her personal
health booklet (described above).
These visits to the Mosoriot Health Center
were critical for establishing a rapport with
the health center’s staff and enlisting their
support. At the same time, directors of the
COBES program at Moi University were in-
volved in supporting the MMRS project as
the data collected in the health center from
the patient care process could then be linked
to the public health data (e.g. description of
households, risk factors for selected infec-
tious diseases) that are collected by medical
students during their household interviews.
4. Overcoming barriers to establishing an
electronic medical record system
None of the staff at the health center have
prior experience with storing information in,
or retrieving information from, computers or
other electronic systems. It is essential that
any system designed to meet the health cen-
ter’s needs be simple in construct, easy to use
by the health center and its support staff, and
easy to maintain and modify once the initial
development and implementation has
occurred.
The first barrier to overcome is electrical
power. Although the Mosoriot Health Center
24 T.J. Hannan et al. / International Journal of Medical Informatics 60 (2000) 21–28
is on the local electrical grid, there are fre-
quent power interruptions and voltage fluctu-
ations. Therefore, secondary power sources
(uninterruptible power sources and a small
generator) and surge protectors are critical to
the survival and ongoing use of the computer
hardware supporting the MMRS.
The second barrier to establishing the
MMRS is familiarity with computers.
MUFHS has substantial computer resources
for the developing world, including student
and faculty computer laboratories. The Dean
of the MUFHS allocated two microcomput-
ers to be placed at Mosoriot for training.
Training of the Mosoriot staff (mostly the
registration clerks who will mainly be using
the MMRS) will be performed by members
of the Indiana University faculty and Moi
University medical students who are trained
to be computer ‘super-users’.
The third barrier to establishing the
MMRS is the lack of a unique patient iden-
tifier. Kenyan society has no national number
such as the social security number in the US.
Therefore, we had to decide, based on the
knowledge of our Kenyan colleagues of
Kenyan society and culture, which set of
variables would most parsimoniously
uniquely identify each patient. After long dis-
cussion, we decided to include the following
fields in the registration process to assign a
MMRS record number (with check digit) to
each patient: the patient’s first name (usually
English), middle name (usually Swahili) and
last name (usually of African origin); the
patient’s birth date; the patient’s mother’s
first name; and the patient’s home village.
(Although in many cases Kenyan adults do
not know their exact date of birth, such a
date is required on a national identity card
they must obtain when they become 18 years
of age. Therefore, they pick a birthdate
which, even if not exact, is consistently held
by the patient.) These same fields will be
collected by the COBES surveys so that the
public health data collected therein can be
linked to clinical care delivered at the Moso-
riot Health Center.
The fourth and greatest hurdle to establish-
ing a functioning electronic medical record
system is making the transition from paper to
the electronic medium. The Mosoriot Health
Center currently has no facilities for elec-
tronic data capture; instead, as described
above it has a redundant system where the
same few pieces of information are collected
and recorded on paper at multiple clinics.
Therefore, the system for data acquisition
must be simple. Electronic data entry must
occur only once and the system must encour-
age accuracy. The foundation of each suc-
cessful electronic medical record system is an
effective data dictionary [3]. It must reflect
both the clinical realities of the care delivered
and serve the administrative and reporting
needs of the providers. A sample of the core
data dictionary for the MMRS is shown in
Fig. 1. It was built for the MMRS by the
authors in face-to-face meetings in Kenya
over a 3-day time frame utilizing the experi-
ence of two authors (WMT, TJH) who have
worked on larger electronic medical record
systems in the US and Australia [4,5]. The
resultant dictionary contains records for se-
lected data elements that are currently
recorded in the health center, with a heavy
emphasis on those required for reporting to
the Kenyan Ministry of Health. We have also
entered clinical diagnoses from all records to
the adult medicine clinic for November 1999,
in order to further populate the data dic-
tionary with records for clinical diagnostic
terms. The fields in the data dictionary in-
clude term number, term name, term type,
term system (e.g. body system, such as cardi-
ology or gastroenterology, or clinical system,
such as laboratory or radiology), term de-
scription, International Classification of Dis-
 T.J. Hannan et al. / International Journal of Medical Informatics 60 (2000) 21–28
eases, Version 10, code and item cost (if
relevant). The other two tables in this rela-
tional database system are the registration
table, containing one record per patient (its
fields being the unique identifiers described
above along with the registration date), and
the visit table, containing one record per visit
(its fields include the patient identifier; the
date and time of check-in; the clinic(s) visited;
ancillary services provided, such as labora-
tory, radiology and pharmacy; charges for
the above services and/or items; the amount
paid; and the time of check-out).
Accurate entry of clinical data is critical to
an electronic medical record system such as
that being created at Mosoriot. Paper-based
encounter forms can serve as data templates
for capturing and entering these data. Even
in places with sophisticated electronic medi-
cal record systems, such encounter forms are
often used [3–6]. Such an encounter form
could also help bridge the gap between paper
Fig. 1. Data dictionary for the Mosoriot medical record system.
25
and electronic media for nurses and staff who
might be fearful that the computer would
radically change their jobs or, worse, replace
them. The Mosoriot encounter form de-
scribed above has been drafted (Fig. 2) and
will be printed on folded cards that, once
used to enter data into the MMRS, will be
given to the patient in lieu of writing infor-
mation into his or her personal health care
booklet. In order to gradually move towards
the electronic patient record, this encounter
form will be used for capturing data 3–6
months prior to installing the MMRS. Data
from these forms will be back-entered into
the system which will not only populate the
MMRS with useful data at its start but will
also allow pilot testing of the data entry
screens and procedures. The data will be
back-entered by the same clerks who will
eventually be entering the data prospectively.
In this way, we can train them to enter data
from the encounter forms and set into mo-
26 T.J. Hannan et al. / International Journal of Medical Informatics 60 (2000) 21–28
Fig. 2. Proposed encounter form for collecting clinical data during patient visits to the Mosoriot Health Center.
tion the checks for transcription accuracy.
The data dictionaries and encounter form
have been integrated into the core electronic
medical record system by two of the authors
(JKR and FS) at Indiana University using
the commercial software Microsoft Access®.
5. System security and functionality
As part of the implementation process, the
MMRS must contain adequate security to
the data and patient information yet fit
within the cost restraints of the Mosoriot
Health Center. Initially, the program will run
on a single standalone microcomputer. It is
envisioned that security will be as follows:
1. Access to the system for all personnel —
developers and users — will be by user-
defined passwords.
2. System security will vary by task and be
assigned only to those users who require
access to each part of the system, e.g.
clinical data entry, dictionary mainte-
nance, program development, etc.
3. Backup of data will be password pro-
tected and performed daily to a Zip drive.
(US$200 plus US$25 per 250 megabyte
capacity disk).
4. Weekly backup of the system and data
will be password protected on a writable
CD-ROM. (US$250 plus 600 megabyte
capacity disks at two for US$25). This
will be stored off site in a secure location
at MUFHS.
5. An uninterruptible power source (UPS)
apparatus and alternative power source
(i.e. a small generator) will provide ade-
quate power. Surge protectors will be
used on all electrical connections.
6. All copies of the completed encounter
forms will be retained initially as a paper
backup to the electronic record during the
3–6 month period when the encounter
forms are used prior to installation of the
MMRS.
 T.J. Hannan et al. / International Journal of Medical Informatics 60 (2000) 21–28
7. During the above 3–6 month period, pa-
tients will have information written in
their booklets, as is the current practice
for the health center. When the MMRS
‘goes live’ and data entry is in real time,
the encounter forms will be given to the
patients for permanent keeping in order
to replace the booklets. Each patient will
be given a small manila envelope in which
to carry the encounter forms.
8. Each patient will have a unique identifier
of six numbers with a check digit that will
be recorded within the record and, in the
future, on the encounter form as a bar
code. This MMRS registration number
and its bar code will also be written on a
label that will be placed on the front of
the patient’s encounter form envelope.
The bar code will increase the speed and
decrease errors of data entry.
9. Using MMRS data for research that is
beyond the usual functions of the Moso-
riot Health Center (e.g. linking clinical
outcomes data with the COBES data col-
lected during household interviews) will
require authorization by the Moi Univer-
sity’s Ethics Committee (its version of the
NIH Institutional Review Board that
must approve all research) in order to
protect patient confidentiality and assure
the appropriate use of these data.
6. Time frame for implementation
Having established the model for the
MMRS, it is envisioned that the time frame
for development will be 6 months and on-site
implementation as a working computer-based
medical record at Mosoriot will take an addi-
tional 3 – 4 months. As stated above, during
the 6-month development period, the encoun-
ter form will be used (once approved by the
Medical Officer and Director of Medical
27
Records of the health center) to collect clini-
cal data prior to installing the record system.
7. Conclusions and insights
Critical to the success of the MMRS pro-
ject is the initial and ongoing involvement of
the clinical and academic stakeholders in the
system at all levels. They have been associ-
ated with the design planning, software devel-
opment and plans for the future evolution of
this electronic medical record system. The
patients remain closely involved in the use of
their medical information by retaining their
accessibility to the individual paper records.
Experience with installing electronic medical
record systems in the US [4] and Australia [5]
have allowed the authors fit a model system
to the Mosoriot Health Center that is likely
to be successful and yet serve the clinical and
research missions of both the Mosoriot
Health Center and Moi University. This ex-
perience has led to the rapid development of
the registration system, data dictionary, visit
database and encounter forms likely to be
clinically useful in the near term, while under-
girding an electronic medical record system
capable of evolving as the local clinical and
research needs change. Similar outcomes
have been seen with the IAIMS projects,
although at levels of magnitude of greater
sophistication [6]. The MMRS experience
reflects the similarities rather than dissimilari-
ties between diverse electronic patient record
systems.
In an economically and technologically de-
prived society, any developments in clinical
information management must meet the local
needs and be sustainable by local resources.
Applying a ‘big bucks’ approach would not
be sustainable by Kenyans once external
funding ends, as it always does. Initially, the
entire MMRS will require only a single se-
28 T.J. Hannan et al. / International Journal of Medical Informatics 60 (2000) 21–28
cure microcomputer, printer, uninterruptible
power source and affordable data security
systems such as bar codes, Zip drives and
CD-ROMs.
The 12-year-old Indiana University–Moi
University collaboration and its 5-year NIH-
funded medical informatics fellowship pro-
gram will lead to the institutional
independence of both the Mosoriot Health
Center and Moi University. This fulfills one
of the aims of the Indiana–Moi University
collaboration, as embodied in a maxim that
is common to both Kenya and the US soci-
eties: ‘‘Give a man a fish, and you feed him
for a day. Teach him to fish, and you feed
him for a lifetime.’’
The facilities for system maintenance and
development must reside among the Kenyan
faculty (initially with JKR, director of this
project in Kenya). Hence, new developments
and individualization of the system to meet
the clinical and research needs of the Moso-
riot Health Center and Moi University can
occur when required. The history of system
development for electronic medical records
and other types of clinical information sys-
tems are sometimes characterized by poor
software support and very little adaptability
to an institution’s individual needs [7]. Such
errors must be avoided if the MMRS and the
clinical and research enterprises which it
serves, are to survive and grow.
Acknowledgements
The authors wish to thank Mr Kimitei, the
Clinical Officer in Charge, and the staff of
Mosoriot Health Center and members of the
Moi University Faculty for the Health Sci-
ences for their support of this project and the
Regenstrief-Moi Medical Informatics Fellow-
ship. We also thank the managers of the
Indiana University-Moi University collabora-
tion for their logistic and intellectual support
of this project. This work was supported by
grant number 1-D43-TW01082 from the Na-
tional Institutes of Health through the Foga-
rty International Center. The opinions are
solely those of the authors and do not neces-
sarily represent the opinions of the NIH or
the authors’ home institutions.
References
[1] R.M. Einterz, C.R. Kelley, J.J. Mamlin, D.E. Van
Renken, Partnerships in international health. The
Indiana University – Moi University experience, In-
fect. Dis. Clin. North Am. 9 (2) (1995) 453 – 455.
[2] R.M. Einterz, R.S. Dittus, J.J. Mamlin, General
internal medicine and technologically less devel-
oped countries, J. Gen. Intern. Med. 5 (5) (1990)
427 – 430.
[3] S.B. Johnson, Generic modelling for clinical reposi-
tories, J. Am. Med. Inform. Assoc. 3 (1996) 328–
339.
[4] C.J. McDonald, J.M. Overhage, W.M. Tierney,
P.R. Dexter, D.K. Martin, J.G. Suico, et al., The
Regenstrief medical record system: A quarter cen-
tury experience, Int. J. Med. Informat. 54 (1999)
225 – 253.
[5] T.J. Hannan, International transfer of the Johns
Hopkins Oncology Center clinical information sys-
tem, MD Comput. 11 (1994) 92 – 100.
[6] W.W. Stead, R. Borden, J. Bourne, D. Giuse, D.R.
Ing, N. Giuse, T.R. Harris, R.A. Miller, A.J. Olsen,
The Vanderbuilt University fast track to IAIMS:
Transition form planning to implementation, J.
Am. Med. Inform. Assoc. 3 (1996) 308 – 317.
[7] W. Slack, Cybermedicine: How computing empow-
ers doctors and patients for better health care,
Jossey – Bass, San Francisco, 1997.