Adding Clinical Utility To The Laboratory Reports Automation Of Interpretative Comments
1. Clin Chem Lab Med 2018; aop
Opinion Paper
Wytze Oosterhuis*
Adding clinical utility to the laboratory reports:
automation of interpretative comments
https://doi.org/10.1515/cclm-2018-0623
Received June 14, 2018; accepted September 20, 2018
Abstract: In laboratory medicine, consultation by add-
ing interpretative comments to reports has long been
recognized as one of the activities that help to improve
patient treatment outcomes and strengthen the position
of our profession. Interpretation and understanding of
laboratory test results might in some cases considerably
be enhanced by adding test when considered appropriate
by the laboratory specialist – an activity that was named
reflective testing. With patient material available at this
stage, this might considerably improve the diagnostic effi-
ciency. The need and value of these forms of consultation
have been proven by a diversity of studies. Both general
practitioners and medical specialists have been shown to
value interpretative comments. Other forms of consulta-
tion are emerging: in this time of patient empowerment
and shared decision making, reporting of laboratory
results to patients will be common. Patients have in gen-
eral little understanding of these results, and consultation
of patients could add a new dimension to the service of
the laboratory. These developments have been recognized
by the European Federation of Clinical Chemistry and
Laboratory Medicine, which has established the working
group on Patient Focused Laboratory Medicine for work
on the matter. Providing proper interpretative comments
is, however, labor intensive because harmonization is nec-
essary to maintain quality between individual specialists.
In present-day high-volume laboratories, there are few
options on how to generate high-quality, patient-specific
comments for all the relevant results without overwhelm-
ing the laboratory specialists. Automation and applica-
tion of expert systems could be a solution, and systems
have been developed that could ease this task.
Keywords: consultation; patient empowerment; reflective
testing.
Introduction
Laboratory testing is the single highest-volume medical
activity. It has been claimed that two-thirds of clinical
decisions are based on laboratory test information [1–3].
Although this claim may be too high, the value of labora-
tory medicine in patient care is undisputed [4]. The core
business of the clinical laboratory is to provide results
of tests requested by physicians and other health care
workers, whereas the task of the laboratory can be defined
in broader terms – to participate in solving diagnostic
challenges. What are the possibilities to add diagnostic
value to laboratory results? What further options can the
laboratories have to improve consultations and support
physicians and patients?
Consultation
Looking back at the developments in the last decennia,
the importance that is attached to consultation is closely
related to the changing views on the position of clinical
chemistry in health care. In the 50s and 60s, the distance
between laboratory and clinic was smaller than is now.
The increasing automation and the associated increase
in the number of tests, and almost completely automated
reporting of results, distanced the laboratory specialist
from the clinic [5, 6]. The increasing scale and emphasis
on the analytical process resulted in large laboratories
mainly focused on a fast response time, but with inher-
ently less attention for clinical support. Automation, cost
savings and streamlining of analytical techniques, with
the reduced focus on supporting the clinical process, has
been identified as a threat to clinical chemistry, even to
the point that it was suggested that clinical chemistry may
not survive [7]. Others were more optimistic, arguing that
we needed to move quickly to meet new challenges and
to seize the opportunities in the preanalytical and the
postanalytical phases, meaning in consultation, with a
need for better integration of the clinic and laboratory [8,
9]. Already in 1996 two independent reports appeared that
*Corresponding author: Wytze Oosterhuis, Department of Clinical
Chemistry and Hematology, EFLM–WG Patient Focused Laboratory
Medicine, Zuyderland Medical Center, Heerlen, The Netherlands,
E-mail: w.oosterhuis@zuyderland.nl
2. 2 Oosterhuis: Automation of interpretative comments
favored this integration [10, 11]. The laboratory should
not be seen as a “number factory”. In a recent review, it
was argued that some tasks of the extra-analytical phase
should become primarily the responsibility of laborato-
ries, including individualized interpretative commenting
[12]. Presently, ISO 15189 states explicitly: “The labora-
tory shall establish arrangements for communicating with
users on the following: advising on choice of examina-
tions and use of the services, including required type of
sample, clinical indications and limitations of exami-
nation procedures and the frequency of requesting the
examination; advising on individual clinical cases; pro-
fessional judgments on the interpretation of the results
of examinations”. This all points to the same conclusion:
consultation is regarded as a central task and competence
of the laboratory specialist. In the above, the problem
to be solved by the laboratory specialist is: how should
one organize the consultation service in these times high
throughput laboratories?
What is the consultation we can do?
Questions concerning individual cases are part of the
daily routine, and in that sense, consultation is a task that
is inherent to the work of every laboratory specialist. What
we are addressing here is consultation – adding special-
ist knowledge – on a larger scale, and what means do we
have to influence the pre- and postanalytical phase?
With regard to the preanalytical phase, most labora-
tories will apply simple rules in defined cases to add addi-
tional tests when appropriate. This is called reflex testing:
a predetermined test protocol is automatically completed.
Examples are the addition of free thyroxin (T4) when
thyroid-stimulating hormone is abnormal, triglycerides
in lipemic samples or bilirubin in icteric samples. More
extensive protocols are being used, such as in anemia
diagnosis [13]. In some cases, a comment may be added
automatically to the test results.
The laboratory specialist might interpret abnormal
test results personally, take other available (medical)
information into account (e.g. age, gender, previous labo-
ratory test results and clinical information) and determine
whether additional tests are indicated. In most cases,
these tests may be performed with the patient’s material
already available in the laboratory. Comments can also
be added to the report to serve the requesting physician.
This process has been called “reflective testing” [14, 15].
The term reflective testing was chosen because this activ-
ity is based on the clinical judgment (reflection) of a
laboratory specialist regarding the interpretation of labo-
ratory results. In this way, laboratory professionals could
add value over the purely analytical service using their
specialist knowledge. It is no exception that in a labora-
tory examination of a patient, abnormal results may be
found that could indicate some unexpected pathology.
Recognition and interpretation of pathological results by
the laboratory specialist may be helpful for physicians
and patients. Examples of disorders typically recogniz-
able by distinct laboratory findings are hemochromatosis,
m-proteins, hyperparathyroidism, vitamin B12 deficiency,
thalassemia, hepatitis, pituitary dysfunction or Gilbert’s
syndrome [16].
A recent Best Practice Report of the Association for
Clinical Biochemistry (ACB) [17] states the following on
reflective testing:
“There can be circumstances where the result of a test or group
of tests will suggest that further investigations should be made
to provide a clearer interpretation or confirm a diagnosis in a
patient.
Best practice: When the reflective test has obvious relevance
to the initial test(s) requested and/or to the medical condition
being investigated or diagnosed then the additional tests can
be performed without necessarily contacting the requestor or
patient. However, this general principle might first need to be
agreed with the service commissioners and users”.
This practice is however laborious and needs harmoniza-
tion between specialists to maintain a comparable quality
level [18].
What is the need for adding tests
and comments?
Several studies have been published on the opinions on
and effects of reflective testing. Some studies focus on
the opinion of the general practitioners or other clini-
cians [19], whereas other studies prioritized the patient’s
perspective [20]. Overall, reflective testing was judged by
physicians as a useful way to improve the process of diag-
nosing (and treating) patients. It has also been shown that
patients will value and support this activity [20].
To study the opinion on reflective testing, 10 clinical
scenarios were circulated to both specialists and general
practitioners, each involving the possible addition of a
specific test [19]. Response options ranged from adding
further tests, phoning the clinician, adding a comment
or just reporting the results. It was concluded that
3. Oosterhuis: Automation of interpretative comments 3
reflective testing is generally welcomed by the doctors,
with the last option – just reporting the results – as the
least favorable. These results were confirmed in a study
in The Netherlands, where reflective testing was judged
to be useful by general practitioners in almost all of the
presented cases [21, 22]. Another study showed a learning
effect: the results showed a better concordance between
the suspected diagnosis and the actions suggested by the
general practitioners if they were or were not familiar with
reflective testing by their laboratory (50.8% vs. 38.2%)
[23]. It was concluded that reflective testing as a form of
consultation can be seen as an added value in the service
of the clinical chemistry laboratory to primary health care
and can also be rewarding for patients.
Interpretative comments for
patients
Apart from adding interpretative comments to reports for
physicians, comments could also be created for patients.
The better-informed patient has been shown to be better
equipped to participate in medical decision processes
[24, 25], contributing to patient empowerment and shared
decision making. Several studies have shown that patients
who are better informed will be better motivated to adhere
to treatment options and that patient empowerment will
result in improved treatment outcomes [26].
Patients express there is a growing demand for better
information in order to participate more actively in treat-
ment decisions. More and more initiatives are being
started to give patients access to their medical records,
often in the form of patient portals, empowering patients
for real participation in diagnosing, treating and monitor-
ing chronic disease.
Although much time, energy and money have been
invested in providing patients with direct access to labo-
ratory test results, this access alone is, however, in general
insufficient for actionable patient knowledge [27]. There is
a fundamental problem of patients not being able to fully
understand their medical data and records. If we burden
patients with the task of figuring out what test results
mean and with the responsibility to act (or not act) based
on that information, then we should take the responsibil-
ity of making these data as meaningful as possible.
The European Federation of Clinical Chemistry and
Laboratory Medicine has recognized these developments.
To this end, the working group Patient Focused Labora-
tory Medicine was established, with the aim to develop
new and direct ways of communication of laboratory
specialists with patients, supporting the role of the lab-
oratory in informing the patients about their test results
and their meaning to them [28].
It should be noted, however, that there is a wide vari-
ation within Europe with respect to patients’ access to
medical records and to laboratory results [29].
In a survey conducted by this working group among
professionals and patients across Europe, it was shown
that there is resistance by some professionals in making
results available directly to patients. In some cases, this
is regulatory, but also because of doubts that patients will
understand the meaning of these laboratory results [30].
With respect to patients, a clear proportion of patients
are interested in receiving their laboratory medicine
results, the majority with explanatory notes; a role for
specialists in laboratory medicine is acceptable to them
and raises the potential for direct engagement by the
laboratory with patients offering a new paradigm for the
provision of laboratory medicine activities [30]. This is a
potential paradigm shift in laboratory relationships with
patients and physicians, and there appears to be an appe-
tite for such progress [27].
Automation of interpretative
comments
Automation of commenting is inevitable if this service is
to be offered on a larger scale, with guarantied continuity
and quality. It should be noted that although reporting to
patients would require different texts than for physicians,
the basic technical solutions will remain the same. It is
expected that the process of the automated generation
of interpretative reports would be considerably improved
when inconclusive laboratory test results are supple-
mented with additional tests. In that way, the options to
be considered for differential diagnosis can be reduced,
and an effective interpretation becomes much more likely.
The automated generation of interpretative comments
will mean the application of expert systems. Although the
history of expert systems in medicine is long, routine appli-
cations are scarce and aimed at confined areas. Several
techniques have been applied [31]: discriminant analysis
was an early technique for the interpretation of sets of data
and was early on believed to find wide application in clini-
cal chemistry [32]. In Bayesian reasoning, the conditional
probabilityofadiagnosisiscalculatedgiventheoccurrence
of the patient variables [33]. Neural networks exist of nodes
and weighted connections, with the nodes distributed
on three main layers: input data, output (diagnosis) and
4. 4 Oosterhuis: Automation of interpretative comments
hidden. Based on historical data, the system will construct
the optimal network, and the larger the amount of data,
the more efficient will be the outcomes.
Although these techniques have and have had some
routine applications, such as in Down syndrome screen-
ing [34], they pose very substantial demands on the con-
struction of the database and harmonization of input data
in order to be transferable [35].
Rule-based systems intend to capture the knowledge
of domain experts into expressions, most often in the form
of if-then statements. A rule-based expert system makes
the storing of large amount of data easy. The rules help
to clarify the logic that is used in the decision-making
process, an advantage over other systems, e.g. neural net-
works, where the reasoning might be obscure.
Current commercial applications in this field are rule-
based systems such as the Gaston system of Medecs [36]
and the RippleDown system of Pacific Knowledge Systems
(PKS) [37].
The Gaston system developed by the ICT Company
Medecs constitutes an advanced knowledge system model
using techniques of artificial intelligence. The Gaston
system is a guideline-based decision support system
[38, 39]. It is currently applied to identify drug-drug inter-
actions [40]. A new application is directed at drug-test
interactions [41].
Ripple-Down Rules (RDR) as applied by the system of
PKS is a general knowledge acquisition technique to build
knowledge-based systems (KBS) incrementally, while the
system is in routine use. It starts with an empty KBS and is
built gradually over time as cases are processed. The lab-
oratory experts build rules as a minor extension to their
normal duties and are able to keep refining rules as KBS
requirements evolve [42].
The two key features of RDR to facilitate adding
knowledge in context are as follows [42]:
– when a conclusion provided by a KBS is incorrect, a
refinement rule is linked to the incorrect rule so that
the refinement rule is only ever evaluated in the same
context, that is, when the parent rule is also valid.
– the expert only ever adds a rule to deal with a par-
ticular case, so that every rule has an associated case
called a cornerstone case. If the expert creates a rule
that is valid not only on the case in hand but also on
other cornerstone cases, they are asked to add condi-
tions to the rule to distinguish the case from the other
cornerstone cases or to accept that the refinement
rule should apply to one of more cornerstone cases.
Commercial systems are now used routinely to provide
detailed interpretative comments to physicians.
It should be recognized that the applications as men-
tioned could find a place in several other related purposes
and fields, including, e.g. intensive care units and appli-
cations for handheld devices. A better use of graphics
and graphical symbols will improve understanding. The
use of international standards, e.g. for nomenclature and
coding of laboratory results, will be necessary for proper
transferability.
It is also a new and interesting development that
larger companies such as Philips and Abbott have recog-
nized the potential of expert system technology to sup-
plement their diagnostic services. Philips and PKS have
signed an agreement to enhance its laboratory informa-
tion management system (LIMS) LABOSYS. Philips has
linked the RippleDown system of PKS to their LIMS. This
promises to offer the possibility to use this system to add
comments (and tests). Abbott started the AlinIQ clinical
decision support initiative, which includes the coopera-
tion with PKS to enhance patient-specific interpretation of
laboratory test results [43].
Conclusions
There is a wide consensus among specialists in clinical
chemistry that consultation – adding value to test results
– is not only an opportunity but also a prerequisite for the
successful development and future of our field. Studies
have shown that specialists, general practitioners and
patients alike wish better information and support in the
interpretation of laboratory tests. If the laboratory is to
fulfil this role, a paradigm shift is needed in reporting of
the results, including interpretative support to patients.
However, we are challenged to find a solution to the
problem of fulfilling this task of adding meaningful inter-
pretative comments (and tests where appropriate) while
dealing with the almost overwhelming numbers of test
results in the present-day laboratory. The only solution
lays in the application of improved information techno-
logy and of expert systems. Applications of this kind are
already available for use on a small scale. However, sub-
stantially more research and developments are needed for
the introduction of such systems on the scale needed.
Author contributions: The author has accepted responsi-
bility for the entire content of this submitted manuscript
and approved submission.
Research funding: None declared.
Employment or leadership: None declared.
Honorarium: None declared.
5. Oosterhuis: Automation of interpretative comments 5
Competing interests: The funding organization(s) played
no role in the study design; in the collection, analysis and
interpretation of data; in the writing of the report; or in the
decision to submit the report for publication.
References
1. Green SF. The cost of poor blood specimen quality and errors in
pre-analytical processes. Clin Biochem 2013;46:1175–9.
2. Plebani M. Exploring the iceberg of errors in laboratory medi-
cine. Clin Chim Acta 2009;404:16–23.
3. Forsman RW. Why is the laboratory an afterthought for managed
care organisations? Clin Chem 1996;42:813–6.
4. Hallworth MJ. The ‘70% claim’: what is the evidence base? Ann
Clin Biochem 2011;48:487–8.
5. Burke MD. Clinical laboratory consultation. Clin Chem
1995;41:1237–40.
6. Burke MD. Clinical laboratory consultation: appropriateness to
laboratory medicine. Clin Chim Acta 2003;333:125–9.
7. Williamson R. Does clinical chemistry have a future? Clin Chem
Lab Med 1998;36:509.
8. McQueen MJ. Evolution or revolution in clinical chemistry. Clin
Chem Lab Med 1999;37:89–90.
9. Dominiczak MH. Laboratory medicine: the need for a broader
view the ‘multiple bundle’ model of clinical laboratory function.
Clin Chem Lab Med 1999;37:97–100.
10. AACC Task Force on the Changing Practice Environment. The
changing environment for the practice of clinical chemistry. Clin
Chem 1996;42:91–5.
11. Athena Society. The future of clinical chemistry and its role
in healthcare: a report of the Athena Society. Clin Chem
1996;42:96–101.
12. Ajzner É. Adding value in the postanalytical phase. EJIFCC
2016;27:166–73.
13. Oosterhuis WP, van der Horst M, van Dongen K, Ulenkate HJ,
Volmer M, Wulkan RW. Prospectieve vergelijking van het
stroomschema voor laboratoriumonderzoek van anemie uit de
NHG-standaard ‘Anemie’ met een eigen, inhoudelijk en logistiek
alternatief stroomschema. [Prospective comparison of the flow
chart for laboratory testing of anemia of the NHG guideline ‘Ane-
mia’ with an substantive and logistical alternative flow chart].
Ned Tijdschr Geneeskd 2007;151:2326–32.
14. Murphy MJ, McMahon MJ, Paterson JR. Reflective testing: the
practice of adding on tests by laboratory staff. Ann Clin Biochem
2005;42:1–2.
15. Paterson JR, Paterson R. Reflective testing: how useful is the
practice of adding on tests by laboratory clinicians? J Clin Pathol
2004;57:273–5.
16. Elnenaei M, Minney D, Clarke DB, Kumar-Misir A, Ali Imran S.
Reflex and reflective testing strategies for early detection of
pituitary dysfunction. J Clin Biochem 2018;54:78–84.
17. Kilpatrick E. Best practice when providing interpretative
comments on laboratory medicine reports (2014). http://acb.
org.uk. Accessed: 15 Oct 2018.
18. Lim EM, Sikaris KA, Gill J, Calleja J, Hickman PE, Beilby J, et al.
Quality assessment of interpretative commenting in clinical
chemistry. Clin Chem 2004;50:632–7.
19. Darby D, Kelly AM. Reflective testing – what do our service users
think? Ann Clin Biochem 2006;43:361–8.
20. Paterson SG, Robson JE, McMahon MJ, Baxter G, Murphy MJ,
Paterson JR. Reflective testing: what do patients think? Ann Clin
Biochem 2006;43:369–71.
21. Oosterhuis WP, Keuren JF, Verboeket-van de Venne WP, Soomers
FL, Stoffers HE, Kleinveld HA. Eigen inbreng van het laborato-
rium – huisartsen positief over ‘reflecterend testen’ [Own input
of the laboratory – GPs positive about ’reflective testing’]. Ned
Tijdschr Geneeskd 2009;153:A486.
22. Verboeket-van de Venne WP, Oosterhuis WP, Keuren JF, Kleinveld
HA. Reflective testing in the Netherlands: usefulness to improve
the diagnostic and therapeutic process in general practice. Ann
Clin Biochem 2009;46:346–7.
23. Verboeket-van de Venne WP, Aakre KM, Watine J, Oosterhuis WP.
Reflective testing: adding value to laboratory testing. Clin Chem
Lab Med 2012;50:1249–52.
24. De Almeida Moura J, Carvalho Costa B, Delbone de Faria R,
Figueiredo Soares T, Perlatto Moura E, Chiappelli F. Improv-
ing communication skill training in patient centered medical
practice for enhancing rational use of laboratory tests: the core
of bioinformation for leveraging stakeholder engagement in
regulatory science. Bioinformation 2013;9:718–20.
25. Cunningham DE, McNab D, Bowie P. Quality and safety issues
highlighted by patients in the handling of laboratory test results
by general practices – a qualitative study. BMC Health Serv Res
2014;14:206.
26. Street RL, Makoul G, Arora NK, Epstein RM. How does
communication heal? Pathways linking clinician – patient
communication to health outcomes. Pat Educat Counsel
2009;74:295–301.
27. O’Kane M, Freedman D, Zikmund-Fisher BJ. Can patients
use test results effectively if they have direct access? BMJ
2015;350:h673.
28. Watson ID. Making test results more easily understood by
patients. BMJ 2015;350:h1942.
29. Watson ID, Siodmiak J, Oosterhuis WP, Corberand J, Jorgensen
PE, Gunnur Dikmen ZG, et al. European views on patients
directly obtaining their laboratory test results. Clin Chem Lab
Med 2015;53:1961–6.
30. Watson ID, Oosterhuis WP, Jorgensen PE, Gunnur Dikmen ZG,
Siodmiak J, Jovicic S, et al. A survey of patients’ views from eight
European countries of interpretive support from Specialists in
Laboratory Medicine. Clin Chem Lab Med 2017;53:1961–6.
31. Al-Badareen AB, Selamat MH, Samat M, Nazira Y, Akkanat O.
A review on clinical decision support systems in healthcare.
J Convergence Inf Technol 2014;9:125–35.
32. Solberg HE. Discriminant analysis in clinical chemistry. Scand J
Clin Lab Invest 1975;35:705–12.
33. Keller H, Gessner U. Bayes’ theorem and quantitative clinical
chemical determination. Clin Chem 1981;27:1959–60.
34. Reynolds TM, Penney MD. The mathematical basis of multi-
variate risk screening: with special reference to screening for
Down’s syndrome associated pregnancy. Ann Clin Biochem
1989;27:452–8.
35. Frølich A, Nielsen BF. Transfer of hypercalcemia discriminant func-
tions between local hospitals. Int J Biomed Comput 1996;41:167–73.
36. Medecs BV. www.medecs.nl. Accessed: 14 Jun 2018.
37. Pacific Knowledge Systems. https://pks.com.au. Accessed:
14 Jun 2018.
6. 6 Oosterhuis: Automation of interpretative comments
38. de Clercq PA, Hasman A, Blom JA, Korsten HH. Design and imple-
mentation of a framework to support the development of clinical
guidelines. Int J Med Inform 2001;64:285–318.
39. Clercq PA. Guideline-based decision support in
medicine. Thesis. The Netherlands: Technische
Universiteit Eindhoven, 2003. ISBN 90-9016967-9.
https://pure.tue.nl/ws/files/2335575/200311945.pdf.
Accessed: 14 Jun 2018.
40. Helmons PJ, Suijkerbuijk BO, Nannan Panday PV, Kosterink JG.
Drug-drug interaction checking assisted by clinical decision
support: a return on investment analysis. J Am Med Inform Assoc
2015;22:764–72.
41. Oosterhuis WP. Personal communication see also: test-medica-
tion interference database. https://www.nvkc.nl/professional/
wat-interfereert-waar. Accessed: 14 Jun 2018.
42. Compton P, Peters L, Edwards G, Lavers TG. Experience with
Ripple-Down Rules. In: Macintosh A, Ellis R, Allen T, editors.
Applications and innovations in intelligent systems XIII. SGAI
2005. London: Springer, 2006.
43. www.abbottdiagnostics.com/AlinIQ. Accessed: 14 Jun 2018.