Original Article

Volume 30, Número 1, p. 64-69, 2014 [Link]

Medical equipment classification according to corrective maintenance

data: a strategy based on the equipment age
Natália Ferreira Oshiyama, Ana Carolina Silveira, Rosana Almada Bassani, José Wilson Magalhães Bassani*

Abstract Introduction: Decision-making on medical equipment management is a daily task for clinical engineers, but
it may prove difficult to easily extract relevant information from the large amount of data from computerized
maintenance management systems. This article describes a simple method of medical equipment classification
based on corrective maintenance indicators. Methods: Three indicators were calculated based on the number
of events, duration and cost of corrective maintenance. Three classes were defined according to the indicator
values of different equipment ages: class A for 0-4 years, class B for 5-9 years, and class C for equipment
older than 10 years. The method was applied to 2,134 pieces of equipment from the Health Service system
of the University of Campinas. Results: From the total, 51.7% of the equipment were classified as C, 4.2%
as B and 44.1% as A. The infusion pump for general use was the type of equipment of which most units
were in the C class (84.7%), even though almost 50% of them were acquired within less than 9 years, and
would thus be expected to be classified as A and B. Among the pumps in class C, 39.5% were from a single
manufacturer, although the equipments were acquired recently. Conclusion: The developed classification
may be an important tool for raising alerts about equipment more prone to maintenance problems, as well
as for identification of equipments with acceptable maintenance history, supporting decision-making on
equipment replacement.
Keywords Clinical engineering, Database, Indicators, Medical equipment classification, Decision-making.

Introduction
The healthcare system has become dependent on Yet, due to the large amount of stored data, relevant
new technologies developed to facilitate patient care. information may not be readily accessible.
According to data from the World Health Organization, Classification may be a useful tool in data analysis.
US$ 8,233 were spent per capita on health in the USA In order to classify items, it is necessary to establish
in 2010 (World ..., 2013), of which it is estimated that some criteria to identify groups (classes) with similar
approximately US$ 412 were destined to medical elements. Once the classes are defined, the items
equipment, assuming that the latter sum corresponds to can be classified according to the description of
5-6.2% of the total health expenditure (Pammolli et al., the attributes of each class (Hastie et al., 2009).
2005). The importance and considerable cost of In this study, equipment age was chosen as the
medical devices emphasize the need of effective criterion for establishment of classes because, with
technology management toward diminishing costs, constant use, replacement of parts tends to occur more
while augmenting quality, efficiency and safety of often, and preventive maintenance may be required
the services provided (Bronzino, 1992). more frequently. This is an important criterion,
since equipment reaches a point in its life-cycle at
A good management practice includes planning,
which frequent failure leads to greater downtime and
acquisition, incoming inspection, inventory, installation, diminished cost-benefit ratio (Clark, 2004). It was
commissioning and acceptance, training of users hypothesized, therefore, that corrective maintenance
and operators, monitoring of use and performance, data (i.e., frequency, duration and cost of services,
maintenance, and replacement or disposal of equipment. which can be easily extracted from the database)
Thus, collecting information on the equipment would provide suitable indicators for the establishment
performance and maintenance constitutes a good of age-based classes, on the assumption that older
basis for planning new acquisitions (Cheng and Dyro, equipment usually fails more often than recently
2004). This can be done by accessing the equipment acquired ones. Equipment was then divided into three
history, usually found in the databases of computerized classes (A, B and C), in which the best maintenance
maintenance management systems (CMMS) available performance, compatible to that of newly-acquired
in most healthcare systems (Cohen and Cram, 2004). equipment, characterizes class A, and the poorest
*e-mail: bassani@[Link]
Received: 16 August 2013 / Accepted: 19 November 2013
Rev. Bras. Eng. Bioméd., v. 30, n. 1, p. 64-69, mar. 2014
Braz. J. Biom. Eng., 30(1), 64-69, Mar. 2014 Medical equipment classification 65

performance characterizes class C. This medical 15 years. Because the age range of the equipment in
equipment classification can help clinical engineers this study was 0-14 years, the maximum depreciation
identify the need for equipment replacement and would be 6.7 % per year.
manufacturer-dependent frequent equipment failure, It has been accepted that, as the equipment gets
as well as the need of training of the maintenance older, failures become more frequent, which increases
group or users. the downtime and the cost of the corrective maintenance
(Bronzino, 1992; Dondelinger, 2003, 2004). Since
Methods this is an intrinsic characteristic of equipment in the
general, it was assumed that age would be a good
A database (Microsoft ExcelTM spreadsheet) containing
information on the equipment (identification number, parameter for establishing limits for the calculation and
location of use, manufacturer, model, acquisition date identification of the maintenance profile. Three classes
and cost of acquisition) and corrective maintenance were then defined according to the profile of equipment
data (number of events, total time and total cost) was of different age ranges: A (0-4 years), B (5-9 years)
produced with data from the CMMS of the Center and C (>10 years). The profile was characterized by
for Biomedical Engineering of the University of the means ( NM , TM , CM ) and respective standard
Campinas, which is responsible for the maintenance of error (SEM) values of the indicators for the three
the medical equipment of the university’s healthcare age ranges, i.e., ( NM ± SEM , TM ± SEM , CM ± SEM ) j ,
area. The data used in this study were collected from where j corresponds to one of the three age ranges.
2004 to 2009. The indicators were compared among classes
From the corrective maintenance data, three with one-way analysis of variance. As it can be seen
primary indicators were calculated: number of in Figure 1, and further explained in the Results
corrective maintenance events (NM), total time section, the initial assumption was supported by
spent on corrective maintenances (TM) and corrective the observation that the values of the indicators
maintenance cost (CM), as follows. significantly increased with the equipment age,
Annual ( Number )i
NM i = (1)
median( Numbert otal )

Annual (Time)i
TM i = (2)
median(Timetotal )

Annual (Cost )i
CM i = (3)
0.06 × ( Acquisition _ Cost )

The index i corresponds to the individual equipment

analyzed (I = 1, 2, …, n), for a total of n equipments.
In order to calculate the indicator NM, the number of
maintenance events for a given item was considered as
the annual average of the six years (Annual(Number)i).
This value was then divided by the median of the
annual number of corrective maintenances performed
by the same maintenance group for all the equipments
in the database (Numbertotal). The latter step makes it
possible to compare the indicator values obtained from
different maintenance groups, which, however, was
not the case here. The same was done to calculate the
indicator for total time of maintenance (TM). The cost
(CM) was calculated by dividing the average annual
cost of corrective maintenance (Annual(Cost)) by 6%
of the equipment acquisition cost. This percentage
is within the expected range of expenditure with
corrective maintenance (3-15% of the acquisition
Figure 1. Mean values and standard error for each equipment age
cost; Cruz et al., 2002) and was calculated by the
range of the corrective maintenances indicators: number of events
straight-line depreciation method (Albrecht et al., (NM, panel A), total time under maintenance (TM, panel B), and
2010) assuming a minimum equipment lifetime of for maintenance cost (CM, panel C).
 Rev. Bras. Eng. Bioméd., v. 30, n.1, p. 64-69, mar. 2014
66 Oshiyama NF, Silveira AC, Bassani RA, Bassani JWM Braz. J. Biom. Eng., 30(1), 64-69, Mar. 2014

indicating that the maintenance performance was the should raise alerts for further investigation, which was
best in class A and the worst in class C. carried on, in this case, by the detailed analysis of
As the standard behavior of the whole equipment four equipment types: 123 physiological monitors, 84
population was identified, individual pieces of general use infusion pumps, 77 pulmonary ventilators,
equipment could be classified according to the values and 50 syringe infusion pumps. These types of
of their indicators and the ranges defined by the equipment are usually located at intensive care units
mean ± SEM of each class. If, however, the value and operating rooms, and they are considered as
of an indicator was between the range limits of two high‑risk devices because their failure might have
classes, the equipment was considered to belong to marked impact on the patients’ well-being (Melendez
the class to which the equipment’s indicator value had and Rane, 2004; Miodownik, 2004).
greater proximity. Classification was determined for Among all equipment, the general use infusion
each of the 3 indicators. The equipment was classified pumps were the type with more items classified as
as C if at least one of its maintenance indicators were C (84.7%). Pulmonary ventilators also showed a
compatible with the C class. On the other hand, the considerable trend to come under class C: 57 out of
condition for the equipment to be classified as A 77 units (74%) (Figure 2).
was that all the 3 indicators were compatible with
Even though almost 50% of the general use
the A class. In all other cases, the equipment was
infusion pumps and pulmonary ventilators were
classified as B. An example of the spreadsheet with
0-9 year-old (Figure 2B), and would thus be expected
the classification of individual indicators and the final
to come under classes A and B, only 15.3% of the
equipment classification is shown in Table 1.
pumps and 23% of the ventilators were classified as
By filtering a particular entry from the spreadsheet
such (Figure 2A). This means that not always does
(e.g. equipment type, model, age, manufacturer,
new equipment need less corrective maintenances,
location of use), scores of each individual piece of
as well as that equipment classified as C may have
equipment may be available for post-hoc analysis,
the maintenance profile similar to that of recently
allowing the clinical engineer to search for specific
acquired devices. This becomes more evident from the
reasons for possibly unexpected results, e.g. score
analysis of general use infusion pumps: for instance,
C for most of the items of a recently acquired batch
all devices from manufacturer M54 were in the age
of devices. The apparently “poor maintenance
range of 0-9 years, yet 83.3% were classified as C
performance” indicated by this score might be due to
(Figure 3). This observation points out a specific
several factors that would require further investigation.
As the number of equipment in the database grows, problem concerning equipment from this manufacturer,
robustness of the analysis increases, as it would be which may be due to the equipment itself, to problems
the limit case of large healthcare conglomerates or a in its usage and/or its maintenance.
nation-wide system.
Discussion
Results The method developed in this study for the classification
As shown in Figure 1, a significant difference between of medical equipment is simple and can be easily
the age ranges was observed for all corrective implemented. It is only necessary to have the corrective
maintenance indicators calculated. The NM values maintenance data from a CMMS and a program
for classes A, B and C were, respectively, 0.40 ± 0.03, that allows the preparation of spreadsheets, such
0.64 ± 0.04 and 0.78 ± 0.06 (F = 21.25; degrees of as Microsoft ExcelTM. In developing countries, it is
freedom = 391, p < 0.001). The values of the TM common the existence of medical equipment with age
indicator were 0.38 ± 0.04, 0.69 ± 0.07 and 0.83 ± 0.09 greater than 10 years, which would be expected to fail
for classes A, B and C, respectively (F = 12.51; more frequently. In this case, the availability of an
p < 0.001), while for CM, the mean values were equipment classification tool to facilitate management
0.24 ± 0.03 for class A, 1.97 ± 0.24 for class B and becomes especially valuable.
2.81 ± 0.42 for class C (F = 31.03; p < 0.001). These In the present method, once the equipment classes
results support the initial assumption of a positive are established based on the average age, any particular
relationship between equipment age and the occurrence device of the same equipment pool is again checked
of corrective maintenance problems. against the value ranges of the indicators for its
From the total of 2134 equipment, 51.7% were individual classification. This means that it may happen
classified as C, 4.2% as B and 44.1% as A. The that an old piece of equipment can be classified as A
elevated number of medical devices classified as C and a new one as C. The greater the equipment pool
Table 1. Data from 36 out of 2134 medical items analyzed in this study, such as the equipment identification (Id), type, location of use (Local), manufacturer, model, serial number (Serial), age (in years) and acquisition
cost (AC; US$), as well as the respective values of the indicators defined in the text (Equations 1-3). The corresponding class of each indicator (NM Class, TM Class and CM Class) and the final equipment classification
(Age Classification) using the method here described are also shown.
Age
Id Type Local Manufact. Model Serial Age AC NM NM Class TM TM Class CM CM Class
Classification
Id1 4 6 12 24 SN1 14 2,612.01 2.5 C 0.65 B 22.95 C C
Id3 7 6 6 12 SN3 14 18,495.90 1.5 C 2.62 C 2.83 C C
Id10 7 1 33 69 SN10 14 104,980.10 0.33 A 0.00 A 0.06 A A
Id13 7 6 24 47 SN13 13 139,799.00 0.67 B 0.41 A 0.40 A B
Id22 7 6 9 17 SN22 12 17,429.40 1.5 C 1.72 C 4.62 C C
Braz. J. Biom. Eng., 30(1), 64-69, Mar. 2014

Id23 4 6 1 1 SN23 12 1,800.00 0.17 A 0.03 A 0.82 A A

Id48 7 1 18 36 SN48 11 9,886.44 0.67 B 0.07 A 0.41 A B
Id52 5 1 22 43 SN52 11 15,526.97 1.17 C 0.65 B 0.39 A C
Rev. Bras. Eng. Bioméd., v. 30, n. 1, p. 64-69, mar. 2014

Id89 4 1 1 2 SN89 10 1,595.60 0.83 C 0.80 C 3.96 C C

Id91 4 1 1 2 SN91 10 1,595.60 1.00 C 0.71 B 4.99 C C
Id92 6 1 7 13 SN92 10 443.34 0.17 A 0.01 A 1.50 B B
Id97 4 6 8 16 SN97 9 1,398.35 0.33 A 0.04 A 1.10 A A
Id118 7 6 6 10 SN118 9 21,941.42 1.17 C 0.79 C 4.08 C C
Id140 1 7 21 41 SN140 9 4,150.48 0.17 A 0.63 B 0.60 A B
Id145 7 6 24 45 SN145 8 11,897.00 0.5 A 0.12 A 0.19 A A
Id166 2 6 19 39 SN166 8 1,815.60 0.83 C 1.7 C 15.59 C C
Id178 3 6 13 25 SN178 7 3,867.25 0.33 A 1.31 C 2.53 C C
Id202 4 1 8 15 SN202 7 5,419.40 0.17 A 0.20 A 0.08 A A
Id205 7 6 31 64 SN205 6 4,977.87 0.33 A 1.99 C 1.32 B C
Id222 1 6 32 67 SN222 5 1,530.85 0.67 B 0.44 A 0.95 A B
Id254 3 6 28 58 SN254 3 12,138.62 0.33 A 0.03 A 0.05 A A
Id258 7 6 18 35 SN258 3 8,538.85 1.00 C 1.32 C 0.74 A C
Id266 4 1 8 15 SN266 2 5,165.86 0,00 A 0.00 A 0.00 A A
Id267 4 1 8 15 SN267 2 5,165.86 0.83 C 3.57 C 0.46 A C
Id268 4 1 8 15 SN268 2 5,165.86 0.5 A 0.09 A 0.15 A A
Id281 4 6 8 15 SN281 1 4,357.31 0.17 A 0.67 B 0.05 A B
Id312 5 6 34 71 SN312 1 25,151.95 1.00 C 1.42 C 0.06 A C
Medical equipment classification
67
 Rev. Bras. Eng. Bioméd., v. 30, n.1, p. 64-69, mar. 2014
68 Oshiyama NF, Silveira AC, Bassani RA, Bassani JWM Braz. J. Biom. Eng., 30(1), 64-69, Mar. 2014

Figure 2. Number of items of four types of high-risk medical equipment in each age range (A) and in each class according to their corrective
maintenance performance (B).

Figure 3. Number of general use infusion pumps from five manufacturers categorized according to the equipment age (A) and to the
classification according to the corrective maintenance performance (B).

used to establish the classes, the more robust is the quantitative data to justify the need to replace the
classification. equipment. But the equipment per se might not be the
With the classification results, many post-hoc only possible cause of failure: error in the equipment
analyses could be made by the professionals responsible use by the clinical staff and/or lack of adequate
for technology management. The four types of maintenance by the clinical engineering group may
equipment further investigated in the present study contribute to it. In the method presented here, the
were chosen due to the high percent of items classified classification according to the corrective maintenance
as C. General use infusion pumps are important life data raises an alert for further investigation of the
support devices used for therapeutic and/or diagnosis possible causes of a maintenance performance not
purposes, but also carry considerable risk in the case compatible with that expected for the equipment age.
of malfunction, as infusion rates different from those Depending on how the information in the spreadsheet
prescribed may be even lethal (Association..., 2010). is filtered, the main cause of the poor classification
Another reason to focus on this type of equipment is could be identified. For instance, if there is a type of
the observation of a typically high number of non- equipment of a given age range and from a certain
programmed service orders, i.e., corrective maintenance manufacturer that is classified as C only at one place
(Taghipour et al., 2010). When the manufacturers of use, this filtering could indicate the necessity of
were compared, M54 was identified as one of the better user training.
sources of a large percent of the pumps classified as It should be stressed that the age-based classification
C (39.5%), even though 94% of the equipment from presented here should not be used as the sole tool for
this manufacturer was 0-4 year-old. Possible reasons support of administrative or regulatory decisions, but
for this atypical maintenance behavior might involve rather as source of alert for the clinical engineering and
the equipment design, parts and/or assembly, as well administrative teams on critical aspects that require
as the clinical and maintenance teams. more detailed investigation. Additional information
It is important to identify individual devices or may be necessary for a correct and robust decision
equipment (and their manufacturers) most prone to (Oshiyama et al., 2012). In the daily, routine use,
failure because they represent poor investment over as it should be the case for all types of indicators,
time, and medical technology managers should have clinical engineers should monitor and investigate
Rev. Bras. Eng. Bioméd., v. 30, n. 1, p. 64-69, mar. 2014
Braz. J. Biom. Eng., 30(1), 64-69, Mar. 2014 Medical equipment classification 69

each aspect, aiming at improving equipment choice PMid:15316975. [Link]

and service delivery, as well as adequately training of 9/50040-5
the health care staff. If used in hospitals nationwide, Cruz AM, Denis ER, Villar MCS, Lic GLM. An event-
this classification might even help raising flags for tree-based mathematical formula for the removal of
regulatory agencies, since these indicators are expected biomedical equipment from a hospital inventory. Journal
to be independent of the size of the hospital or the of Clinical Engineering, 2002; 27(1):63-71. [Link]
org/10.1097/00004669-200202710-00039
size and quality of the maintenance groups.
Dondelinger RM. A simple method of equipment
replacement planning. Biomedical Instrumentation &
Acknowledgements Technology, 2003; 37(6):433-6.
Authors are indebted to the Clinical Engineering and Dondelinger RM. A complex method of equipment
Informatics teams from the Center for Biomedical replacement planning. Biomedical Instrumentation
& Technology, 2004; 38(1):26-31. [Link]
Engineering of the University of Campinas for their org/10.2345/0899-8205(2004)38[26:ACMOER][Link];2
invaluable assistance during collection and analysis of
equipment data. Financial support: Conselho Nacional Hastie T, Tibshirani R, Friedman J. The elements of statistical
learning: data mining, inference and prediction. New York:
de Desenvolvimento Científico e Tecnológico (SAE/
Springer; 2009. [Link]
UNICAMP, CNPq, Grant. N. 831346/1988-4, NFO,
Grant. N. 300632/2005-3, JWMB). Melendez L, Rane R. Anesthesiology. In: Dyro JF, editor.
Clinical engineering handbook. Burlington: Elsevier
Academic Press; 2004. p. 384-91. [Link]
References B978-012226570-9/50099-5

Association for the Advancement of Medical Miodownik S. Operating room. In: Dyro JF, editor. Clinical
Instrumentation - AAMI. Infusing patients safely. Priority engineering handbook. Burlington: Elsevier Academic
issues from the AAMI/FDA infusion device summit. Press; 2004. p. 376-84.
Arlington: AAMI; 2010. 41 p. Oshiyama NF, Bassani RA, D’Ottaviano IML, Bassani
JWM. Medical equipment classification: method and
Albrecht WS, Stice EK, Stice JD. Financial accounting.
decision-making support based on paraconsistent
Ohio: Cengage Learning; 2010.
annotated logic. Medical & Biological Engineering &
Bronzino JD. Management of medical technology - a primer Computing, 2012; 50(4):395-402. PMid:22407498. http://
for clinical engineers. Boston: Butterworth-Heinemann; 1992. [Link]/10.1007/s11517-012-0888-6
PMid:1596745.
Pammolli F, Riccaboni M, Oglialoro C, Magazzini L, Baio
Cheng M, Dyro JF. Good management practice for medical G, Salerno N. Medical devices competitiveness and impact
equipment. In: Dyro JF, editor. Clinical engineering handbook. on public health expenditure. Luxembourg: Office for
Burlington: Elsevier Academic Press; 2004. p. 108-10. http:// Official Publication of the European Communities; 2005.
[Link]/10.1016/B978-012226570-9/50035-1
Taghipour S, Banjevic D, Jardine AKS. Reliability analysis
Clark TJ. Health care technology replacement planning. In: of maintenance data for complex medical devices. Quality
Dyro JF, editor. Clinical engineering handbook. Burlington: and Reliability Engineering International. 2010; 27(1):71-
Elsevier Academic Press; 2004. p. 153-4. [Link] 84. [Link]
org/10.1016/B978-012226570-9/50046-6
World Health Organization - WHO. World Health
Cohen T, Cram N. Computerized maintenance management Statistics 2013 [internet]. 2013 [cited 2013 June 08].
systems. In: Dyro JF, editor. Clinical engineering handbook. Available from: [Link]
Burlington: Elsevier Academic Press; 2004. p. 124-30. world_health_statistics/2013/en/.

Authors
Natália Ferreira Oshiyama, Ana Carolina Silveira, José Wilson Magalhães Bassani*
Departamento de Engenharia Biomédica, Faculdade de Engenharia Elétrica e de Computação, Universidade Estadual de
Campinas – UNICAMP, Rua Alexander Fleming, 105, Cidade Universitária Zeferino Vaz, CEP 13083-881, Campinas, SP,
Brasil.

Rosana Almada Bassani, José Wilson Magalhães Bassani

Centro de Engenharia Biomédica, Universidade Estadual de Campinas – UNICAMP, Campinas, SP, Brasil.