J Med Syst (2012) 36:557–567

DOI 10.1007/s10916-010-9501-4

ORIGINAL PAPER

An Intelligent Healthcare Management System: A New

Approach in Work-order Prioritization for Medical
Equipment Maintenance Requests
Naser Hamdi & Rami Oweis & Hamzeh Abu Zraiq &
Denis Abu Sammour

Received: 16 February 2010 / Accepted: 20 April 2010 / Published online: 4 May 2010
# Springer Science+Business Media, LLC 2010

Abstract The effective maintenance management of medical number of medical devices requiring tracking and manage-
technology influences the quality of care delivered and the ment in a hospital may range from 1,000 devices for
profitability of healthcare facilities. Medical equipment smaller community hospitals to over 10,000 for large,
maintenance in Jordan lacks an objective prioritization academic, medical centers [1]. Wang et al. have indicated
system; consequently, the system is not sensitive to the impact that the most common cause of medical equipment
of equipment downtime on patient morbidity and mortality. downtime is poor maintenance, planning, and management.
The current work presents a novel software system (EQUI- Consequently, they have extensively discussed and
MEDCOMP) that is designed to achieve valuable improve- reviewed medical equipment inclusion criteria, as well as
ments in the maintenance management of medical technology. the application of statistical techniques, in medical equip-
This work-order prioritization model sorts medical mainte- ment management plans [2, 3]. Medical equipment man-
nance requests by calculating a priority index for each request. agement is of particular importance in developing
Model performance was assessed by utilizing maintenance countries, where resources and alternatives are scarce, as
requests from several Jordanian hospitals. The system proved such, the creation of a carefully-designed equipment control
highly efficient in minimizing equipment downtime based on and management system can be of vital importance. This
healthcare delivery capacity, and, consequently, patient can be achieved by employing computerized maintenance
outcome. Additionally, a preventive maintenance optimiza- management systems (CMMSs) as a fundamental informa-
tion module and an equipment quality control system are tion resource, providing the technology management staff
incorporated. The system is, therefore, expected to improve with a wealth of support-related information as well as
the reliability of medical equipment and significantly improve assisting management in decision making [1].
safety and cost-efficiency. Additionally, as medical equipment becomes increasing-
ly more sophisticated and plays a more crucial role in
Keywords Biomedical technology management . modern healthcare, maintenance and management issues
Work-order prioritization . Preventive maintenance . demand ever-increasing attention. Development of CMMSs
Quality control is essential for managers and engineers, not only to provide
quick management solutions, but also to predict future
outcomes based on historical equipment performance data.
Introduction The most commonly employed methods of work-order
prioritization for repair requests in Jordan are variants of the
Safe, effective, and economic use of medical devices within first-come, first-served (FCFS) method. While the FCFS
a hospital requires tracking each individual device. The approach might be acceptable for many applications, it is
not always appropriate when applied to the healthcare
N. Hamdi (*) : R. Oweis : H. Abu Zraiq : D. Abu Sammour sector, as is the case when a vital, life-support machine
Biomedical Engineering Department,
undergoes failure and, consequently, is out of service until
Jordan University of Science and Technology,
Irbid, Jordan the service work-order reaches the head of the queue. One
e-mail: [email protected] approach to address these shortcomings requires that
558 J Med Syst (2012) 36:557–567

hospitals maintain a full-time equipment oversight commit- performance statistics database increases [20, 21]. While
tee to prioritize all maintenance requests. Unfortunately this such models represent an advanced tool for quality control
is a time- and cost-intensive approach. Another approach measurements, it is highly sophisticated and computation-
suggests focusing on the risk posed by equipment failure on ally complex, as such a simpler model requiring a smaller
larger groups of patients, rather than focusing on the data set would be beneficial.
equipment with the highest maintenance demand [4]. Some There are numerous models that deal with PM optimi-
authors have suggested categorizing systems according to zation, work-order prioritization, and quality control in the
their level of complexity as a guide for system manage- industrial sector, but little work has focused on the
ment, optimization, and cost reduction, [5] as well as application of these models to the healthcare sector. As a
proposing a rule base for real-time equipment replacement result, many of the existing models focus on minimizing the
prioritization [6]. While various commercial computerized cost of service delivered rather than the human cost of
maintenance management systems are available, [7] there is equipment downtime and failure. Consequently, this work
little objective published work available. The model aims to develop a system that addresses these issues with
presented herein relies on an intelligent work-order prior- particular stress on human cost, well-being and safety.
itization system. It enables the medical service provider to
construct a real-time prioritized equipment service list for
submitted maintenance requests based on various prede- Material and methods
fined factors such as equipment function, location of use,
time since request was issued, availability and distance to This study presents an intelligent medical equipment
nearest substitute, and the overall rate of equipment utilization. management system named EQUIMEDCOMP. The system
Industrial maintenance management information tech- was programmed using Microsoft Visual Basic (version 6).
nology systems have been in use for many years. The The overall flow chart of EQUIMEDCOMP is illustrated in
research has addressed equipment classification systems, Fig. 1a and b, which show the system’s various modules,
preventive maintenance (PM) scheduling models, and tools, equations, and databases, as well as the relationship
work-order systems for prioritizing repair requests for between them.
industrial facilities and manufacturing companies, [8–12] EQUIMEDCOMP can be utilized to carry out multiple
but very few of these systems have addressed the specific medical equipment management-related tasks. The system
needs of the healthcare management field. Commonly was designed in a modular format to allow independent
accepted maintenance policies include age-replacement development of the individual modules (subsystems). These
PM, [13] as well as the periodic PM and sequential PM subsystems include medical equipment inventory, work-
policies [14]. In the periodic PM policy, devices subject to order system, PM scheduling system, and equipment
degradation are maintained in fixed predefined time quality control system.
intervals, independent of machine failure rate, while in a
sequential PM policy the PM time intervals become shorter Medical equipment inventory subsystem
and shorter as time passes (i.e. more frequent PM is
required as the device ages). Other studies, such as those The first and the most critical step in implementing an
carried out by Badia et al. [15] Berenguer et al. [16] and equipment management system (computerized or non-
Yang et al. [17] attempted to optimize PM periods by computerized) is to build an extensive inventory of all the
relying on the continuous assessment of the equipment equipment to be tracked by the program. Without an
condition and attempting to predict the level of perfor- effective inventory system, it is impossible to track
mance degradation. We propose a model adapted from the equipment-management functions accurately.
work of Adzakpa et al. [18] that utilizes an optimization The proposed system is supplied with an extensive
algorithm for PM periods and combines it with a fixed medical equipment inventory that continuously tracks all
periodic PM approach, to yield higher accuracy and stability. services being performed on each medical device. Each
Incorporating a quality control module into the CMMS device has an equipment control number (ID); the control
can provide objective, quantitative, and reliable assessment number is assigned automatically by the system during
of equipment performance. Such approaches include those device registration, and is used for identification within the
based on the quality function deployment (QFD) approach system. In addition to the ID number, the system records
[19]. This approach allows service to be tailored towards basic technical information, such as the serial number,
the actual demand, and is characterized by a “semi- device name, type, manufacturer, model, location of use,
quantitative” and objective approach to quality assessment manufacture date, acquisition date and cost, installation
in healthcare structures. Generally the accuracy of assessing date, equipment function, and the number of preventive
a medical device’s reliability increases as the size of the maintenance tasks specified by the manufacturer per year.
J Med Syst (2012) 36:557–567 559

Fig. 1 a: EQUIMEDCOMP
system flow chart demonstrat-
ing the functionality of the
equipment inventory subsystem.
The various commands (shown
as icons) and their interconnec-
tions with databases or tables
(shown as cylinders) are dem-
onstrated. b: EQUIMEDCOMP
system flow chart demonstrating
the work-order prioritization,
PM scheduling, and quality
control subsystems
560 J Med Syst (2012) 36:557–567

All data, except for the ID number, is acquired during an Table 1 The values assigned to equipment by function and location
equipment registration procedure through a web portal. The Equipment function (EF)
well organized and user-friendly environment is designed to EF Numeric value
facilitate and speed-up the process of data entry. Therapeutic - Life-support 10
The system is also equipped with a complete hospital Therapeutic - Surgical or Intensive Care 9
database which includes basic information about hospitals in Therapeutic - Physical Therapy or Treatment 8
contract with the service-providing agency. This information
Diagnostic - Surgical or Intensive Care Monitoring 7
may include the hospital name, address, city, zip code, GPS
Diagnostic - Other physiological monitoring 6
coordinates, phone number, fax, website, e-mail, number of
Analytical - Laboratory 5
beds, and bed occupancy percentage, among other facts.
Analytical - Computer and related 3
A third database includes records of the company’s
Miscellaneous - Patient-related 2
service employees and their basic information. This feature
Miscellaneous - Non-patient related 1
provides a means of communication between employers
Location of equipment use (L)
and employees, allows for easier job allocation between
L Numeric value
biomedical equipment technicians, and holds records of
Anesthetizing Locations 5
employees’ work activity.
Critical Care Areas, Operational Rooms 4
The system is equipped with searching options that can
Wet Locations/Labs/Exam Areas 3
be used to find any device within the system’s inventory.
General Patient Care Areas 2
The search process could be initiated according to one or
Non-Patient Care Areas 1
more of the following: Equipment ID, hospital name,
equipment type, and equipment name.

Work-order subsystem for prioritizing maintenance requests

An important feature that is incorporated into our system is Table 2 The values assigned to equipment based on hospital load
(size), time since service was requested, and distance to nearest
a work-ordering subsystem that assigns a priority number
alternative of the device in question
for each unscheduled maintenance request. This subsystem
documents incoming requests for maintenance services and Hospital Load (number of beds)
keeps track of the work-order until completion. The priority HL Numeric Value HL Numeric Value
number for any request represents a calculated numeric >550 12 251–300 6
value indicating the relative importance of that request. It 501–550 11 201–250 5
enables the system to qualify a certain request to be more 451–500 10 151–200 4
important than another based on its medical necessity and 401–450 9 101–150 3
patient safety. 351–400 8 51–100 2
Upon entering the system, a maintenance request is 301–350 7 0–50 1
subjected to a special testing algorithm consisting of six Time ( in days)
factors, each of which assigns a certain numeric value to T Numeric Value T Numeric Value
that request. These factors test the importance of the request >10 22 5 10
based on the following criteria: (1) equipment function, (2) 10 20 4 8
location of use, (3) the load on the hospital containing the 9 18 3 6
failed device, (4) the presence of an alternative to this 8 16 2 4
device in the hospital, (5) time since maintenance request in 7 14 1 2
days, and (6) distance to the nearest hospital containing the 6 12
same type of device for which maintenance is requested. Distance to nearest alternative ( km)
The priority number of each request is then calculated as a D Numeric Value D Numeric Value
weighted sum of six different numeric values developed by >90 26 20.1–30 12
each factor, and the device with the highest priority number 80.1–90 24 10.1–20 10
is serviced first. Numerical weights assigned to the various 70.1–80 22 5.1–10 8
factors were determined by conducting recursive iterations 60.1–70 20 2.1–5 6
of the prioritization algorithm towards optimal work-order 50.1–60 18 1.1–2 4
sequences, as deemed by the expert opinions of local 40.1–50 16 0–1 2
physicians and clinical engineers; these values may be
30.1–40 14
found in Tables 1 and 2.
J Med Syst (2012) 36:557–567 561

Any equipment registered in the system is assigned an A in Eq. 2 a numeric value of 10 for equipment without any
equipment function (EF) number indicating the functional readily available substitute, and setting the value of A to 0 if
category of the device, as seen in Table 1. In this regard, an alternate is available. The time since the maintenance
therapeutic life-support equipment is considered more request was issued (T) is also considered in the system (see
important to a patient’s life than equipment in other Table 2), since equipment requiring maintenance is consid-
functional categories, and thus receives a higher numeric ered out of service, thus increasing equipment downtime.
value. In other words, the most crucial equipment receives Based on these six factors, namely, EF, L, HL, A, T, and
the highest numeric value and, as a result, if it fails it D, each device requiring maintenance is given a priority
receives the highest priority number. number; the higher the priority number the higher the
Another requirement of equipment registration in the necessity for maintenance. The priority number, P, is
system is to specify the location in which the equipment is calculated according to Eq. 2:
used; consequently a numeric value (L) is given for each
device indicating the area in which it is primarily used. The P ¼ ðX1 ÞEF þ ðX2 ÞL þ ðX3 ÞA þ ðX4 ÞT þ ðX5 ÞD
system sorts equipment into five categories, as indicated in þ ðX6 ÞHL ð2Þ
Table 1. Devices used in operating rooms (OR) and critical
care areas receive a higher numeric value (greater priority) where X1, X2, X3, X4, X5, and X6 are the weights assigned to
as their failure represents an imminent threat to a patient’s EF, L, A, T, D, and HL, respectively.
life. The weight assigned to each factor when computing the
Since bed occupancy reflects the percentage of the total priority number was determined based on recursive converg-
number of beds that are occupied by patients at any given ing iterations to minimize the prioritization error between
time, it may be used to assess utilization of the facility. The expert prioritization (physicians and clinical engineers) and
product of hospital capacity (given by the number of beds that of the computerized system (EQUIMEDCOMP). The
within the hospital) and bed occupancy may be used as an optimum case, which generated the best outputs (least match
indicator of the hospital load (HL). As HL increases, the error), was to assign an approximate weight of 0.35, 0.35, 0.1,
number of patients expected to need a certain device 0.1, 0.05, and 0.05 to X1, X2, X3, X4, X5, and X6., respectively.
increases, and, as such, the system gives this device a Although these assigned weights produced efficient and
higher priority number. Numeric values assigned to realistic outputs in our tests, the administrator has the ability
different ranges of HL are indicated in Table 2. to adjust these values according to his/her professional
The geographical location of the hospital and the judgment and the particular institutional requirements, see
distance to the nearest hospital with an appropriate Fig. 2. As expected, equipment function (EF) and location of
substitute for the failed device is accounted for within the use (L) received the highest weight, followed by the presence
system. If a device in a certain hospital fails, the system of an alternative to the failed device within the same hospital
locates the nearest registered hospital within the system (A) and the time since the maintenance request was issued
database containing an alternate device. This is done with (T), and finally the distance to the nearest alternative (D) and
the assistance of a virtual global map, for example Google the hospital load (HL).
Earth, and employing the spatial coordinates of the two The data required for the operation of the work-order
hospitals according to Eq. 1: subsystem is gathered in different stages: information regard-
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ing the device location, type, and fault description is entered
ð x1  x2 Þ 2 þ ð y1  y 2 Þ 2 through the “Add fault” window, seen in Fig. 3. Information
D¼ ð1Þ regarding equipment function, location of use, and the
1000
presence of an alternative within the hospital is assigned
Where D is the distance (km) to the nearest hospital when the equipment is registered in the system, as seen in
containing a similar functioning device, (x1,y1) and (x2,y2) Fig. 4. The system uses the GPS coordinates of a hospital to
are the spatial coordinates of the hospital containing the calculate the minimum distance between it and all other
failed device and the nearest alternate, respectively. The registered hospitals containing an available alternate.
equation was divided by 1,000 to express D in kilometers.
The system then assigns a higher priority to the device Preventive maintenance scheduling subsystem
whose alternate is the farthest away, as shown in Table 2.
The presence of an alternative to a device is an important Medical equipment failure is an extremely sensitive issue,
issue to consider. Medical equipment without any substitute since every medical device is directly and closely related to
within the same hospital should have preference when the patients’ health and well being. Thorough PM can
maintenance is required. The system assigns equipment with virtually eliminate downtime caused by equipment failure,
no substitute or alternate a higher priority, by giving variable but the associated cost of such maintenance can be prohibi-
562 J Med Syst (2012) 36:557–567

Fig. 2 The EQUIMEDCOMP

priority table of devices for
which maintenance has been
requested and prioritized

tive. Therefore, a compromise must be found that minimizes which is a measure of how frequently a medical device
the cost associated with periodic inspections and extends the undergoes failure, expressed in failures per hour.
durability of medical equipment. It is clear that a properly The failure rate of a device per year (λ) can be obtained by
weighted PM strategy is necessary for medical equipment to dividing the total number of failures per year by the cumulative
operate in the safest and most cost-effective manner. annual operating time of the device as seen in Eq. 4:
EQUIMEDCOMP incorporates a preventative mainte- n
nance scheduling sub-system that assigns an optimum l¼ ð4Þ
h
period between inspections, based on an algorithm adapted
from a period optimization approach and the classical fixed where n is the total number of failures for all devices of
period approach. the same type and same manufacturer per year, and h is
From the installation time of the device, and assuming the cumulative annual operating hours for all devices of the
exponential behavior, the availability of the device at time t same type and manufacturer.
is given by Eq. 3 [18]: Assuming that almost all medical devices are repairable,
the failure rate of a device is set equal to the repair rate (i.e.
m l λ = μ), hence Eq. 3 simplifies to:
AðtÞ ¼ þ eðlþmÞt ð3Þ
mþl lþm
AðtÞ ¼ 0:5 þ 0:5e2mt ð5Þ
where μ is the repair rate of the device, which is a measure
of how frequently a medical device is repaired, expressed in
repairs per hour; and λ is the failure rate of the device,

Fig. 3 The “Add fault” window for requesting service for a device
registered within the system Fig. 4 EQUIMEDCOMP equipment registration window
J Med Syst (2012) 36:557–567 563

Clearly, the availability of a device is an exponentially system that can perform reliability assessment of medical
time-decaying function. Consequently, if no maintenance is technology to ensure safe and cost-efficient functionality is of
performed on the device, its availability diminishes toward a vital importance. The outcomes of this technology assessment
threshold value of ω, and whenever A(t) ≤ ω the device is can help in planning and deploying future technology
considered unavailable [18]. Accordingly, on account of the requirements, determining whether a particular medical
constraint that the availability A(t) should always exceed the system needs to be replaced, and helping in brand selection
threshold value ω, the following inequality is derived: of the purchased technology. Usually, the quality evaluation
(control) process takes the form of statistical surveys that are
AðtÞ  w ð6Þ
performed during the life span of a particular device. These
Substituting for A(t) from expression (3) into (6), the surveys may include various parameters, such as the device’s
latter becomes probability of failure, mean downtime, usability, and perfor-
mance measures. These statistical surveys give insight into the
0:5 þ 0:5e2mt  w ð7Þ actual performance of medical equipment.
From which it may be inferred that The equipment quality control sub-system employed by
  EQUIMEDCOMP provides an objective and quantitative
1 reliability assessment for the registered devices. It aids in
t  lnð2w  1Þ ð8Þ
2m the identification of equipment that should be replaced, as
well as decisions on the subsequent brand selection for new
The preceding model represents the period-optimizing
devices being purchased. The sub-system includes an
approach for PM scheduling. EQUIMEDCOMP incorporates
integrated history record (time-line) that keeps track of all
this, together with the fixed-periodic approach that is based on
maintenance and repair activities performed on a particular
scheduled preventive maintenance periods assigned by the
device throughout its service life.
manufacturer, to ensure the consideration of the manufac-
The system is capable of carrying out accurate calcu-
turer’s recommendations and at the same time accounting for
lations of equipment service life for any selected period of
the state of the device with time, from which Eq. 9 is deduced: time by generating a quality report, Fig. 5. This report
   
1 1 contains a set of parameters that give a quantitative
t PM ¼ þ  lnð2w  1Þ  365 ð9Þ
x 2m indication of equipment performance. The most frequently
used parameters include the probability of failure (Eq. 12),
where x is the scheduled preventive maintenance rate per the average number of failures per year (Eq. 13), the failure
year and ω is probability of failure-free work, as represented ratio per year relative to the total number of medical
in Eq. 10 devices (Eqs. 14 and 15), and the mean service life before
first failure (Eq. 16). These equations were adapted from
N  nyearly
w¼ ð10Þ the work of Dori et al. [19] and Toporkov, [20] and are
N given in terms of the time fraction (Tf) as defined in Eq. 11:
where N is the total number of medical devices in the same
ðDn Þ
category and made by the same manufacturer as the device Tf ¼ ð11Þ
for which PM is required; nyearly is the total number of 365
failures per year occurring in the device requiring PM; and where Dn represents the number of days of the time period
τPM is the time in days at which the PM should be performed over which the quality report is requested, specified by a
from t=0. start and end date that are entered by the user.
If corrective maintenance is carried out on the device at t = T, The probability of failure (f) is given by
then the next PM will be scheduled after τPM, i.e. at t = T + n
τPM. The system calculates τPM only for devices in which PM f ¼ ð12Þ
N  Tf
is specified when the equipment is registered in the system.
where n is the total number of failures within the same category
Equipment quality control subsystem and of the same manufacturer of a medical device, and N is the
total number of medical devices of the same category and of
The quality of healthcare delivered in modern medical the same manufacturer registered within the system.
facilities is directly dependant on the state of the medical The average number of failures per year ðnÞ is calculated
technology and equipment employed. As such, in order to from
provide patients with quality medical care, healthcare insti-
n
tutes continually strive to enhance the quality of the medical n¼ ð13Þ
technology they employ. Developing a technology evaluation Tf
564 J Med Syst (2012) 36:557–567

Fig. 5 Quality report window

for registered devices

The failure ratio per year relative to the total number of The mean service life before first failure (tmean) is
medical devices of the same manufacturer (FRsm) is given expressed as:
as
P
N
n Sl
FRsm ¼ ð14Þ i¼1
N tmean ¼ ð16Þ
N
While the same ratio across manufacturers (FRac) is
given by where Sl is the service life of all medical devices before the
occurrence of the first failure.
n
FRac ¼ ð15Þ Other useful parameters employed are the mean correc-
Nac tive maintenance time (Eq. 17), the mean downtime per
where Nac denotes the total number of devices across year (Eq. 18), and the mean time between failures. These
manufacturers. equations were adapted from the work of Amari et al., [21]

Table 3 Maintenance requests submitted to a third-party medical service provider within a four-day work period. The time each order was
received rounded to the nearest 5 min is indicated

Requests 1st Day 2nd Day 3rd Day 4th Day

1 Blood Cell Counter-Al Bashir Dialysis Unit-Prince Zaid EEG Monitor-Prince Refrigerated Centrifuge-
Hospital-9:00 AM Hospital-8:00 AM Hashem-8:30 AM Palestine Hospital- 9:15 AM
2 Anesthetic Ventilator-Prince Coagulation Analyzer- Defibrillator-Prince Ali EMG monitor-Private
Hashem Hospital-9:30 AM KAUH- 9:00 AM Hospital-9:15 AM Clinic- 11:30 AM
3 Electrical Bed-Jordan Medical Ventilator-Jordan CT-Prince Ali Hospital- Medical Refrigerator-JUH-
Hospital-10:00 AM Hospital-9:20 AM 10:00 AM 01:45 PM
4 Ultrasound Imager -Princess Therapeutic Ultrasound Defibrillator-Prince Zaid
Haya Hospital-11:00 AM unit- JUH- 10:10 AM Hospital-11:00 AM
5 Infusion Pump-Princess Basma MRI-KAUH-11:00 AM Lithotripsy Machine-Jordan
Hospital-12:00 PM Hospital-12:00 PM
6 UV imager-KAUH-12:30 AM X-Ray-Princess Basma Electrosurgical Unit-JUH-
Hospital-11:30 AM 1:00 PM
7 Spectrophotometer-JUH- ECG Monitor-JUH-12:00 PM Dialysis Unit -Princess Haya
1:00 PM Hospital-2:00 PM
8 CT-Prince Ali Hospital MRI-Prince Ali Hospital-
Hospital-2:00 PM 1:00 PM
9 Doppler Ultrasound Defibrillator –Ramtha
Machine-Ramtha Hospital- 1:30 PM
Hospital-2:10 PM
J Med Syst (2012) 36:557–567 565

Table 4 Maintenance requests for the four-day period sorted using the FCFS method

Request 1st Day 2nd Day 3rd Day 4th Day

1 Blood Cell Counter-Al UV imager-KAUH Coagulation Analyzer-KAUH ECG Monitor-JUH

Bashir Hospital
2 Anesthetic Ventilator-Prince Spectrophotometer-JUH Medical Ventilator-Jordan MRI-Prince Ali
Hashem Hospital Hospital Hospital
3 Electrical Bed-Jordan Hospital CT-Prince Ali Hospital Therapeutic Ultrasound Defibrillator –Ramtha
unit-Prince Ali Hospital Hospital
4 Ultrasound Imager -Princess Doppler Ultrasound Machine- MRI-KAUH EEG Monitor-Prince
Haya Hospital Ramtha Hospital Hashem
5 Infusion Pump-Princess Dialysis Unit-Prince Zaid X-Ray-Princess Basma Defibrillator-Prince
Basma Hospital Hospital Hospital Ali Hospital

where the mean corrective maintenance time (MCMT) is The overall function of the equipment quality control
given by the following expression: sub-system is to generate a quality report containing the
above mentioned parameters for any given medical device,
P
n
li ðTcorr Þi manufacturer, or combination of both within a specific time
i¼1
MCMT ¼ P
n ð17Þ frame, Fig. 5.
li
i¼1

Such that Tcorr is the total time duration of corrective Results and discussion
maintenance.
The mean downtime per year (T D ) is given by: The work-order prioritization system (EQUIMEDCOMP)
was simulated and tested against variants of the long-used
Pn   P n
TD i
TDi FCFS approach, such as that employed by the Biomedical
Tf
Directorate (BMD) in Jordan. Service requests to a third-
T D ¼ i¼1 n ¼ i¼1 ð18Þ
Tf n party medical service provider were used to assess system
performance; for demonstration purposes, the results of the
where TDi = Ti + (Tcorr)i first 4 days of testing (28 service requests) are presented in
The mean time between failures (MTBF) in hours is Table 3. For comparison and testing, requests were
given as the inverse of the failure rate (λ), and the mean processed using both the FCFS and EQUIMEDCOMP
time to response (MTTR) is given by: models. Based on a survey of local medical maintenance
P
n service providers it was found that an average service
li Ti provider could respond to a maximum of five maintenance
i¼1
MTTR ¼ Pn ð19Þ requests per day, thus any excess maintenance requests are
li postponed to the following day. The work schedule for the
i¼1
4-day period using the FCFS method is shown in Table 4.

Table 5 Maintenance requests for the four-day period sorted using EQUIMEDCOMP. The priority number (PN) for each request is indicated

Requests 1st Day 2nd Day 3rd Day 4th Day

1 Anesthetic Ventilator-Prince Therapeutic Ultrasound Dialysis Unit-Princess Haya Medical Ventilator-Jordan

Hashem Hospital PN=6.7 unit- JUH Hospital PN=6.45 Hospital PN=7.5 Hospital PN=5.8
2 CT-Prince Ali Hospital MRI-Prince Ali Hospital Electrosurgical Unit-JUH CT-Prince Ali Hospital
PN=6.15 PN=6.15 PN=6.45 PN=5.5
3 Infusion Pump-Princess Basma MRI-KAUH PN=6.1 Lithotripsy Machine-Jordan X-ray- Princess Basma
Hospital PN=5.25 Hospital PN=6.1 Hospital PN=5.05
4 Ultrasound Imager -Princess Dialysis Unit-Prince Zaid Defibrillator-Prince Ali UV imager- KAUH
Haya Hospital PN=4.65 Hospital PN=5.9 Hospital PN=6.05 PN=4.55
5 Doppler Ultrasound Machine- Defibrillator –Ramtha Defibrillator-Prince Zaid Blood cell counter - Al-Bashir
Ramtha Hospital PN=3.75 Hospital PN=5.5 Hospital PN=6.0 Hospital PN=4.3
566 J Med Syst (2012) 36:557–567

The EQUIMEDCOMP system prioritized the mainte- equipment reliability and availability by assigning shorter
nance requests as previously outlined, resulting in a PM time periods in response to increased failure rate of a
drastically different work schedule when compared to the particular device.
FCFS method. Factors such as equipment function, location Finally, the quality control sub-system is an excellent
of use, hospital load, time since maintenance request, tool for aiding management and engineering staff in
presence of an alternative, and distance to nearest hospital decision-making by providing a wealth of information
containing the same device had a notable impact on the regarding registered medical equipment reliability and
prioritization of maintenance requests. Table 5 illustrates performance indices. As an example, the historical perfor-
the work schedule determined by EQUIMEDCOMP mance data for a ventilator was used to generate the 2009
according to the priority number calculated for each of the (calendar year) quality report for the device, as shown in
requests. Fig. 5. The quality report reveals the device’s performance
It is important to note that postponed requests received a indices for the indicated period, such as, the probability of
correspondingly higher priority number the following day. failure, number of failures, downtime per year, and mean
For example, the blood cells counter at Al-Bashir Hospital service life before first failure. It will require several years
initially received a priority number of 3.7. Accordingly, the of performance data accumulation before the full potential
device was not listed on the work-order list on the first day. of this module may be assessed.
On the second day, that same device achieved a priority
number of 3.9, which was still not high enough for it to be
included on the work-order list. On the third day, the blood Conclusion
cell counter achieved a priority number of 4.1 which, again,
didn’t qualify it to be serviced due to the fact that more The EQUIMEDCOMP work-order prioritization system for
important requests were introduced that day. However, on medical equipment maintenance management proved its
the 4th day the device received a score of 4.3, which was effectiveness in prioritizing maintenance requests. More-
sufficient to list and service the device. The same applies to over, it demonstrated sensitivity to patient safety and
the maintenance request received for the medical ventilator healthcare quality. The model also aided in more efficient
at Jordan Hospital which was received on the 2nd day and scheduling of PM actions and was responsive to the actual
was postponed until the 4th day according to its relative performance of equipment. This, along with quality control
importance when compared to other requests. indicators, can increase reliability and availability of
By comparison, using the FCFS variant, many important medical equipment. The system can serve as a comprehen-
high-priority requests were postponed to the following day, sive medical maintenance management system for medical
including a CT at Prince Ali Hospital, a defibrillator at service providers (third-party service providers) and med-
Prince Zaid Hospital, and a lithotripsy machine at Jordan ical service departments of hospitals and governmental
Hospital. Using the EQUIMEDCOMP work-order prioriti- healthcare institutions giving quantitative solutions to
zation system resulted in the more important requests management problems and ensuring safe and cost-
(greater impact on patient outcome) receiving higher effective operation.
priority, and, consequently, being serviced more rapidly.
Additionally, the system’s adaptability, with regard to
providing operator flexibility and control over the priority References
equation weight allocation, was greatly appreciated by the
companies asked to evaluate the system, since it creates a 1. Dyro, J., Equipment control and asset management, computerized
more dynamic program. maintenance management systems. Clin. Eng. Handb. 1:122–130,
With regard to the PM scheduling sub-unit, a sample of 2004.
2. Wang, B., and Levenson, A., Equipment inclusion criteria—a new
50 medical devices of the same type and manufacturer were interpretation of JCAHO’s medical equipment management
studied. These devices operate normally 9 h per day, standard. J. Clin. Eng. 25(1):26–35, 2000.
312 days per year, resulting in a total of 2,808 h of 3. Wang, B., and Rice, W. P., JCAHO’s equipment inclusion criteria
operation per year. The average failure rate of the devices revisited—application of statistical sampling technique. J. Clin.
Eng. 28(1):37–48, 2003.
was calculated to be 3.205×10−3 failures per hour or 8.99 4. Wang, B., Furst, E., Cohen, T., Keil, O. R., Ridgway, M., and
failures per year (Eq. 13). The probability of failure-free Stiefel, R., Medical equipment management strategies. Biomed.
work, ω, was then found to be 0.82 (Eq. 10), and the Instrum. Technol. 40(3):233–237, 2006.
resulting PM time period, τPM, was calculated to be 5. Youssef, N. F., and Hyman, W. A., A medical device complexity
model: a new approach to medical equipment management. J.
175.8 days or 5.78 months (Eq. 9). The PM period Clin. Eng. 34(2):94–98, 2009.
scheduling model was sensitive to failure rates of equip- 6. Taylor, K., and Jackson, S., A medical equipment replacement
ment, which, upon implementation, enhanced medical score system. J. Clin. Eng. 30(1):37–41, 2005.
J Med Syst (2012) 36:557–567 567

7. Cram, N., Computerized maintenance management systems: a 15. Badía, F. G., and Berrade, M. D., Optimum maintenance of a
review of available products. J. Clin. Eng. 23(3):169–179, 1998. system under two types of failure. Int. J. Mater. Struct. Reliab. 4
8. Joo, S.-J., Scheduling preventive maintenance for modular (1):27–37, 2006.
designed components: a dynamic approach. Eur. J. Oper. Res. 16. Berenguer, C., Chu, C., and Grall, A., Inspection and maintenance
192:512–520, 2009. planning: an application of semi-Markov decision process. J.
9. Grigoriev, A., van de Klundert, J., and Spieksma, F. C. R., Intell. Manuf. 8:467–476, 1997.
Modeling and solving the periodic maintenance problem. Eur. J. 17. Yang, Z., Djurdjanovic, D., and Ni, J., Maintenance scheduling in
Oper. Res. 172:783–797, 2006. manufacturing systems based on predicted machine degradation.
10. Pongpech, J., and Murthy, D., Optimal periodic preventive J. Intell. Manuf. 19:87–98, 2008.
maintenance policy for leased equipment. Reliab. Eng. Syst. Saf. 18. Adzakpa, K. P., Adjallah, K. H., and Yalaoui, F., On-line
91:772–777, 2006. maintenance job scheduling and assignment to resources in
11. Percy, D. F., and Kobbacy, K. A. H., Determining economical distributed systems by heuristic-based optimization. J. Intell.
maintenance intervals. Int. J. Prod. Econ. 67:87–94, 2000. Manuf. 15:131–140, 2004.
12. Chareonsuk, C., Nagarur, N., and Tabucanon, M. T., A multi- 19. Dori, F., Iadanza, E., Bottacci, D., and Mattei, S., A QFD-based
criteria approach to the selection of preventive maintenance approach to quality measurement in health care. IFMBE Proc.
intervals. Int. J. Prod. Econ. 49:55–64, 1997. 16:1102–1106, 2007.
13. Badia, F. G., Berrade, M. D., and Campos, C. A., Optimal 20. Toporkov, A. A., Criteria and methods and methods for assessing
inspection and preventive maintenance of units with revealed reliability of medical equipment. Biomed. Eng. 42(1):11–16, 2008.
and unrevealed failures. Reliab. Eng. Syst. Safety 78:157–163, 21. Amari, S. V., Zuo, M. J., and Dill, G., O(kn) Algorithms for
2002. Analyzing Repairable and Non-repairable K-out-of-n:G Systems.
14. Wang, H., A survey of maintenance policies of deteriorating Handbook of Performability Engineering (Misra, K., ed.). Chapter
systems. Eur. J. Oper. Res. 139:469–489, 2002. 21: 309–320. Springer: London, 2008.