Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

Laboratory

automation and
information
management
ELSEVIER Laboratory Automation and Information Management 33 ( 1998) 163- 168

LIMS: An automating or informating technology?


J.E.H. Stafford *
Motherwell Information Systems, Sension House, Denton Drive, Northwich, Cheshire CW9 7LlJ, UK

Abstract

The hypothesis that automation of laboratory processes through LIMS implementation is not itself sufficient to greatly
benefit or provide strategic advantage to an organisation is discussed. Traditionally automation has been implemented in
order to increase management control over the means of production. As a result the introduction of automation into much of
the manufacturing industry has led to increased job dissatisfaction, worker alienation and industrial unrest. This is the case
irrespective of the automating technology, e.g. steam-powered machines or computer controlled systems. The specifiers and
implementors of LIMS have happily trodden down this well-worn path, without digesting the experiences of other industrial
sectors. However, the use of Information Technology to automate business processes provides a unique, additional
dimension. IT can provide information about the automated process itself. In this respect IT is an informating technology.
The use of computers within an informating strategy focuses on opportunities for continual learning as new data, events and
contexts characterising the automated process create chances for additional insights, improvements and innovation. LIMS
must informate as well as automate in order to achieve a successful implementation. 0 1998 Elsevier Science B.V. All rights
reserved.

Keywords: Laboratory automation; Informating strategy; User satisfaction; Data-centric: LIMS; Process-centric

1. Introduction vendors in addition to their implementation partners


in Europe and USA alone.
Computers have found widespread use within lab-
Nevertheless, in spite of the rapid adoption of
oratories. Initially used to facilitate management of
LIMS, particularly within regulated industries, there
data from increasingly more automated analytical
is much anecdotal and some published information
equipment, computers were subsequently employed
that suggests the user satisfaction level for installed
to automate the potentially error-prone clerical activ-
systems is around 50% [I]. The considered response
ities associated with the collation of sample and test
to this observation was to increase the perception of
data and subsequent reporting to the client. The
LIMS ownership by involving the user earlier in the
various software applications developed to meet these
decision making stages and to increase the project
automated clerical requirements were classified un-
and management skills to the level necessary for
der the category of laboratory information manage-
successful project completion. Whilst these initia-
ment systems, LIMS. Commercialization was rapid
tives may well facilitate the selection, specification
and the LIMS business now supports at least 80
and successful implementation of a standard LIMS
there is no a priori evidence to suggest that user
* Tel.: +44-l-606330010: e-mail: jstafford@mother.co.uk. satisfaction rate will increase as a result.

0925-5281/98/$ - see front matter 0 1998 Elsevier Science B.V. All rights reserved.
PII: S1381-141X(97)00017-8
164 J.E.H. Stafford/Laboratory Automation and Information Management 33 (1998) 163-168

Is it possible that the user’s and industry’s percep- (laboratory) business practices at the expense of
tion of LIMS requirements has become fixed in the increased user dissatisfaction.
past and prevents the development and implementa- The increased use of computers and mechanical
tion of more user-acceptable LIMS software applica- devices to automate portions or all of routine analyti-
tions? If so, what has been overlooked and how can cal procedures resulted in an explosion of paper-work
we exploit this insight to improve the effectiveness because the analysis of the instrumental response
of LIMS applications and implementation? was no longer the rate limiting step for data collation
Computers are unique in their capacity to auto- [4]. The rate limiting step became the effort required
mate in that they are also capable of providing to manually transfer data from document to docu-
additional information about the underlying auto- ment as it passed through subsequent data collation,
mated process. This property of computer systems manipulation and approval processes prior to genera-
was recognized by Zuboff [2], who coined the word tion of the final report or certificate and filing.
‘informate’ to describe it. Users can exploit the Transcription errors were common and required
informating properties of computer-based systems to careful review processes to detect them.
facilitate their understanding of the automated pro- There was also an increased awareness of the
cess e.g. analytical process, sample collection and importance of quality issues and quality assurance
preparation systems, data management. Data charac- programs were implemented not only to comply with
terizing these automated processes e.g. system suit- regulatory guidelines e.g. GLP and GMP, but also
ability, QC results, analyte recoveries, sample back- for commercial reasons. These programs also added
log and turnaround times are readily generated, but to the laboratory workload and need for improved
need to be turned into information to be of practical documentation management procedures.
use to aid decision making. The resulting increased Given this background it is perhaps not surprising
potential for control over the automated processes that laboratory managers sanctioned the rapid take
will lead to improved job satisfaction since the user up of a technology, LIMS, that provided a means of
will no longer feel that they are at the mercy of, or reducing the need for data transcription and automat-
threatened by, the computer. They will become mas- ing the reviewing and reporting of data.
ter of the computer, not continue as its perceived A laboratory can be viewed as a high level pro-
slave. Consequently, more emphasis must be placed cess wherein samples and their accompanying paper-
on laboratory processes and applications that create work are inputs and the outputs are analytical data in
information from procedural data acquired and man- the form of reports [5]. A measure of laboratory
aged by a LIMS in order to empower the LIMS user. performance is productivity in terms of the number
New LIMS should redress the imbalance between of samples analyzed and error-free reports issued in
data and processes and become more process-centric. a given accounting period. Thus business reasons
alone were sufficient to justify using computers to
automate the data collation, manipulation and report-
2. Why is automation not sufficient? ing processes, thereby increasing productivity. How-
ever, experience in other sectors suggested that this
There are a number of ways of defining automa- approach to automation leads to general worker dis-
tion [3] with computers providing a mechanism by satisfaction for those who must interact with the
which a process can be made automatic. In defining computerized process [2]. The driving force for au-
an automation strategy for the laboratory, LIMS was tomation in this scenario is the laboratory business
classified under information management together manager’s need to control the means of production
with, for example, databases and spectral libraries. rather than facilitating the analyst’s work.
This is consistent with the widespread perception
2.1. Impact of automation
that the sole purpose of LIMS is to automate labora-
tory clerical activities. However, only the status quo In an analysis of the application of computers to
[2] is maintained when computer based systems are automate non-laboratory activities in other industrial
used in this manner to facilitate improvements in sectors Zuboff [2] came to number of pertinent con-
J.E.H. Stafford/Laboratory Automation and Information Management 33 (1998) 163-168 165

elusions. Firstly the desire to automate was driven by Changes in requirements that reflect new reports
the manager’s need to exert more control over the or data entities for management are readily managed
means of production. In doing so the manager by modem LIMS with data base configuration and
usurped any element of control that the worker had reporting tools. However, changes in business rules
over their own work activities, thus demotivating the or processes cannot, because they are almost always
workforce. Secondly, worker motivation was not embedded deep within the LIMS executable or
affected when automation was introduced that facili- database management system. Consequently data
tated participation of the workforce in the decision base driven, data-centric LIMS will always tend to
making necessary to maintain the automated process. maintain the status quo.
Thirdly, automation maintains the status quo.
There is an uncanny parallel between the first
observation described by Zuboff and experiences
described by laboratory users of LIMS. 3. Informating properties are not exploited in
Managers of laboratories within larger organiza- LIMS
tions are also affected. There are many good Corpo-
rate reasons for wanting to make analytical data Successful use of computers to automate business
readily available to groups outside the laboratory for practices must exploit the informating properties of
further manipulation or to support decision making. computerized systems. Standard LIMS do not exploit
However, the ‘fire and forget’ reporting model for informating properties very well. A good example is
laboratory automation does not fit into this new the mandatory audit function for LIMS in regulatory
vision where external user groups require a filtered environments.
view of the analytical database [Sl. If external users The use of computer systems to automate labora-
are able to interrogate the database on demand, tory clerical activities created an additional challenge
bypassing laboratory generated reports, the manager in that there was an urgent need to be able to
may resent this loss of control over a perceived demonstrate the accuracy and integrity of electroni-
product of the laboratory process. This could lead to cally stored data. Regulatory guidelines required that
local resistance to the use of computer systems to corrected data in paper systems remain legible, but
implement a Corporate wide data-sharing system. standard computer systems over-wrote the data item
The third observation has an interesting implica- when edited.
tion for LIMS applications and their implementation. Data management systems in multi-user environ-
All published advice for those intending to imple- ments automatically record the time and date and
ment a LIMS advocates modeling and understanding nature of access to a record and the user ID. How-
the current laboratory practices and then modeling ever, the data is transient in that only the most recent
the desired state. In principle this should reflect an access is recorded. Where there is need to demon-
optimized model, but in practice will reflect the strate data integrity all write and delete access must
perceived capabilities of the yet to be chosen system, be recorded. Consequently an audit application was
particularly in the case of the first implementation specified that would track and record all changes to
[6]. The advice is valid in that a laboratory does not critical data items and store the data in a readily
want to encapsulate inefficient practices. However, retrievable form. Note that the resulting audit appli-
the corollary is that it is very difficult, if not practi- cation could not have been developed were it not for
cally impossible, to readily change a LIMS configu- the appropriate specification of the underlying data
ration once implemented and inefficient, old or re- or file management systems.
dundant practices become embedded in the auto- However, although the audit function informs the
mated system. The ability to quickly respond to user of changes to the data it does not fully exploit
change is a feature of successful business practice the informating properties of computer systems. Ulti-
and a LIMS that cannot readily meet these new mately it is necessary to run a report to extract the
requirements must always fall short of expectation audit data for an object of interest. There is no
and a candidate for replacement [6]. mechanism for assessing the data in order to make a
166 J.E.H. Stafford/Laboratory Automation and Information Management 33 (1998) 163-168

decision concerning what the audit data illustrates those procedural aspects that comprise the total labo-
about the data entry and manipulation processes. ratory activity. Laboratory personnel from the most
Similar considerations apply to object histories junior technician up to the laboratory director own
e.g. chain of custody for a sample. If the LIMS just these processes. They will have a vested interest in
tracks the date, current location of the sample and its ensuring that the work they do, viz. the process they
current owner then it is a data reporting function i.e. are involved with, is of the highest quality. However,
data-centric. The system records what has happened, in order to maintain the processes they need to
not what should happen. measure the performance of the process and charac-
This type of LIMS limitation was recognized and terize the parameters that impact on process perfor-
discussed some years ago in reference to the backlog mance. Employing LIMS to facilitate these processes
report [7]. This is standard report in all LIMS, but it will be of little value unless implemented with a
is just a report. In order to make the reported data view to exploiting the informating properties of com-
useful it must be turned into information vide infra. puter systems.
Consider for a moment chromatography systems
as part of data management automation. In order to
4. LIMS as an informating technology monitor instrument performance a system suitability
test is performed prior to injecting the calibration
An informating strategy for the implementation of standards, samples and quality control samples. From
data management automation will help to redress the a purely data model perspective the system suitabil-
imbalance apparent in the laboratory by focusing ity test will not have the same characteristics as the
more on the processes than on the crude laboratory calibration standards or samples. Consequently a
inputs and outputs. Laboratory processes are owned standard LIMS that generates test worksheets will
by the laboratory personnel who have direct impact not be able to record system suitability unless there
upon the quality of ‘their’ processes and conse- is a mechanism for associating the suitability test and
quently on the quality of the process products (out- results with that particular instance of the instrument
puts) e.g. analytical data, reports. An informating run or test worksheet. The test results may thus not
strategy will help to identify the software applica- be recorded with the appropriate data set or in the
tions necessary to facilitate decisions concerning how LIMS and will not readily be available for query
the automated process is behaving, leading to in- should there be an issue with regard to the analytical
formed corrective actions to maintain or improve results. Likewise it is common practice to only record
process quality. A process-centric approach will also the calculated result in the LIMS. This practice
influence LIMS architecture, which as a result will results in a loss of information because the chro-
become more amenable to changes in business prac- matographic data, retention times and peak area or
tices. Indeed the impact may be as fundamental as height are not stored in a readily retrievable manner.
influencing properties of the objects used to build These data yield valuable information about the ana-
LIMS of the future. lytical process itself, in addition to the QC results.
The historical emphasis placed on the visible An informating strategy requires not only storage of
product of laboratory activities viz. analytical data all the chromatographic data, but also statistical qual-
and reports has naturally led to the development of ity and process control applications to analyze the
data-centric LIMS applications, where more empha- data. These applications should be seamlessly acces-
sis is placed on data management automation than on sible from within the test worksheet to facilitate
management of the automated processes that produce analysis of the analytical process.
the data. Traditional measures of laboratory produc- The requirement to store raw analytical data in an
tivity based on data reported are no longer viable in easily retrievable form is a feature of the LASF
a Corporate vision of data accessible to all with process model for laboratory automation [8] and is
appropriate authority. The laboratory is just another consistent with the implications of adopting an infor-
service function amongst many others. One way for mating strategy.
the laboratory to regain its identity is to refocus on The limitation of using LIMS to generate labora-
J.E.H. Stafford/Laboratory Automation and Information Management 33 (1998) 163-168 167

tory performance data i.e. sample backlog report was An interest possibility arising from the workflow
discussed above. An informating approach implies a paradigm is the potential capability for the system to
requirement to transform sample backlog data into learn. If it were possible to build the workflow on
information by comparing the backlog against a set the fly for an ad hoc sample as the analyst was
of limits viz. a specification is set for a measure of determining the best analytical approach, the suc-
the sample backlog. Bespoke software is required to cessful route could be subsequently stored. The next
implement this concept in standard LIMS as a back- time a similar sample was presented the system
log management system, because the LIMS model could select or suggest the appropriate workflow. If
for specification checking is limited to comparing in addition the system captured the reasoning behind
analytical results against product specifications. the analyst’s route i.e. the business rules, elements of
A characteristic of the data-centric LIMS model is an expert system would be in place.
use of the ‘status’ attribute e.g. available, in testing, A further implication of exploiting informating
rejected, to describe the state of laboratory entities properties of computer systems relates to the con-
e.g. sample, test, result, project. This feature of struction of the LIMS software application itself. The
LIMS is sold by virtue of offering the user the increasing use of object oriented technology in the
possibility of monitoring the workflow. By itself this construction of new LIMS applications provides for
feature is not very useful since the user can only the opportunity to specify that the objects and their
generate reports summarizing laboratory objects of a methods should be self-testing. The recording and
certain status i.e. the current state of the laboratory. storage of the outcomes from the self-testing pro-
What is potentially of more use is to determine what cesses will provide the evidence required to support
will happen. the premise that the systems are operating or con-
LIMS use many of these object statuses to apply tinue to operate in a valid manner i.e. as specified.
laboratory business rules e.g. a rejected result cannot The software driven processes that enable com-
be approved. However, since the rules are usually puter systems to function and communicate are taken
deeply embedded within the architecture of the LIMS very much for granted. Yet the stochastic nature of
it is impossible to implement new a business rule the underlying electronic processes dictates that there
that states, for example, ‘in certain circumstances a will be a significant amount of self-verification built
rejected result can be approved’ without extensive into the system. For example, when data is packaged
customization. An informating strategy expressed in for transfer between computers across a network a
LIMS design would focus on processes and provide check sum is automatically calculated based on the
a tool set that would enable the implementor to data content and included in the frame for transmis-
model the laboratory processes on the screen and sion [9]. The receiving computer unpacks the frame,
define the business rules that determine how the recalculates the check sum and accepts the transmis-
processes link together and transform input into out- sion if the check sums agree. This model would
puts. LIMS would become a workflow application readily translate to a scenario familiar to users carry-
and thus able to meet changing business require- ing out Black Box validation of their LIMS. How-
ments more readily. ever, instead of the users constructing a test protocol
As a workflow application there are many possi- to demonstrate that data entered at a screen prompt
bilities for managing and understanding laboratory was stored in a correct manner, the software would
processes by analysis of data associated with the carry out the check automatically. The software
workflow. For example, if due dates are included ‘methods’ to capture and process screen input and
after each task in the workflow the LIMS could store the data in the database management system
automatically record actual times to complete the would be automatically encoded and compared with
task. Actual times could be used to refine task timing the value previously calculated for a known success-
or predict availability of approved results or flow ful execution of the function (unit or module test). If
bottlenecks. Limits set against critical tasks in the the values agree the data is stored, if not the system
workflow would provide early warning of a break- tries again or reports an error. In this scenario the
down in the workflow. emphasis is on processes not the data. Data integrity
168 J.E.H. Stafford/Laboratory Automation and Information Management 33 (1998) 163-168

and validation is managed by the database manage- tion, Chem. Intell. Lab. Syst. Lab. Inf. Manage. 17 (1992)
265-282.
ment system since the charac-teristics of the data
[4] D.J. Betteridge, Towards intelligent automation, Anal. Proc.
elements comprising the screen prompts are inherited
24 (1987) 106-108.
from the database meta-data definitions. [5] J.E.H. Stafford, LIMS: An Automating or Informating Tech-
nology in Advanced LIMS Technology - Case Studies and
Business Opportunities, J.E.H. Stafford (Ed.), Publ. Blackie
Academic and Professional, London, 1995.
References 161 T.V. Iorns, Replacement LIMS: Moving Forward or Maintain-
ing the status quo. in Advanced LIMS Technology - Case
Studies and Business Opportunities, J.E.H. Stafford (Ed.),
[II R.D. McDowall, A Model for a Comprehensive LIMS in Publ. Blackie Academic and Professional, London, 1995.
Advanced LIMS Technology, Case Studies and Business Op- [7] R.R. Mahaffey, Laboratory Automation in LIMS - Applied
portunities. J. E. H. Stafford @d.l, Publ. Blackie Academic Information Technology for the Laboratory, R.R. Mahaffey
and Professional, London, 1995. (Ed.), Publ. Van Nostrand Reinhold, NY, 1990.
El S. Zuboff, In the Age of the Smart Machine: The Future of 181 J. Liscouski, Laboratory and Scientific Computing A Strategic
Work and Power, Publ. Heinemann Professional Publishing, Approach, Publ. John Wiley and Sons, NY, 1995.
Oxford, UK, 1988. [9] S.P. Maj, Networking Concepts for Chemists, Chem. Intell.
[31 R.D. McDowall, Strategic approaches to laboratory automa- Lab. Syst. Lab. Inf. Manage. 13 (1992) 203-210.

You might also like