Which Why and What Do The Results Mean 2018 Texas Association for Clinical Laboratory Science Thursday 22 Mar 2018 Wyndham El Paso Airport Frank H Wians Jr PhD MTASCP MASCP DABCC FACB ID: 812958
Download The PPT/PDF document "Clinical Laboratory Tests:" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Clinical Laboratory Tests:Which, Why, and What Do The Results Mean?
2018 Texas Association for Clinical Laboratory ScienceThursday, 22 Mar 2018Wyndham, El Paso Airport
Frank H. Wians, Jr., PhD, MT(ASCP), MASCP, DABCC, FACB
Professor of Pathology
Texas
Tech University of the Health Sciences Center El Paso
and the
Paul L. Foster School of Medicine
Technical Director, Clinical Chemistry, University Medical Center (UMC) El Paso
Medical Director, UMC Far East and West Clinic Laboratories
Lt Col (
Ret
), USAF, Biomedical Sciences Corps
Editor-in-Chief
,
Lab
Medicine
, 2004-2011
frank.wians@ttuhsc.edu
Slide2Disclosures and ReferencesDisclosures
NoneSelected ReferencesWians FH Jr. Clinical laboratory tests: which, why, and what do the results mean? Lab Med. 2009;40:105-113.Wians FH Jr, Gill GW. Clinical and anatomic pathology test volume by specialty and subspecialty among high-complexity CLIA-certified laboratories in 2011.
Lab Med
. 2013:44:163-167.
Carraro P, Plebani M. Errors in stat laboratory: types and frequencies 10 years later. Clin Chem. 2007;53:1338-1342.Fillippo A, et al. Artificial neural networks in medical diagnosis. J Appl Biomed. 2013;11:47-58.Obuchowski NA, Lieber ML, Wians FH Jr. ROC curves in Clinical Chemistry: Uses, Misuses, and Possible Solutions. Clin Chem. 2004;50:1118-1125.Ziadie M, Wians FH Jr. A guide to the interpretation of CSF indices. Lab Med. 2005;36:558-562
2009-present
Slide3OutlineThe Laboratory Testing Cycle
Diagnostic Decision MakingMedical NecessityQuestion to Ask Before Ordering a Laboratory TestReasons for Ordering a Laboratory Test
Approaches to Establishing a Diagnosis Based on Laboratory Test Results
Clinical Performance Characteristics of Laboratory Tests
Receiver-operator characteristic (ROC) curvesReference Interval for Interpreting Laboratory Test ResultsCritical Difference Between Consecutive Laboratory Test ResultsArtificial Neural Networks (ANNs)Summary
Slide4Distribution of Test Volume (~6 Billion) by Clinical or Anatomic Pathology Specialty in CLIA-Certified High Complexity Laboratories (n = 172,000 in 2011)*
*Wians FH Jr, Gill GW.
Lab Medicine
. 2013;44:163-67.
Summary (2011)
~6 billion laboratory tests performed annually in ~172,000
high complexity
labs (~72%) out of the 239,000 labs in the U.S.
AP tests represent ~10% of all tests;
CP (laboratory medicine) tests represent the remaining
90
%!
Slide5The Laboratory Testing CycleLaboratory Testing Cycle
Slide6% of Stat Test Errors by Category:Preanalytical (PreA), Analytical (A), and
Postanalytical (PostA)Adapted from: Carraro P, Plebani
M. Errors in stat
laboratory
: types and frequencies 10 years later. Clin Chem. 2007;53:1338-1342.5 PreA types of errors accounted for 44.4% of all PreA errors!If the clinical laboratory is provided with the:Right specimenAt the right
time
On the correct patientWith a complete and accurate test request
The
laboratory
will
provide
a quality result in
a
timely manner.
What did the data look like in another 10 years (2017)?
Slide7Recent Source of Pre-analytical Error:Biotin Interference in Biotin-Streptavidin Immunoassays
Ortho
Vitros
Tests
Affected by Biotin Interference Performed in the UMC Clinical Lab(Quantitative, numerical value; Qualitative, Pos/Neg
)
Test Abbreviation
Type of Result Reported
Effect of Interference
AFP
Quantitative
Falsely
low
CA 125
Quantitative
Falsely
low
CA 15-3
Quantitative
Falsely
low
CEA
Quantitative
Falsely
low
CK-MB
Quantitative
Falsely
low
Ferritin
Quantitative
Falsely
low
FSH
Quantitative
Falsely
low
iPTH
Quantitative
Falsely
low
LH
Quantitative
Falsely
low
Myoglobin
Quantitative
Falsely
low
NT-
proBNP
Quantitative
Falsely
low
Prolactin
Quantitative
Falsely
low
PSA
Quantitative
Falsely
low
Total beta-hCG
Quantitative
Falsely
low
Troponin I
Quantitative
Falsely
low
TSH
Quantitative
Falsely
low
B12
Quantitative
Falsely
high
Cortisol
Quantitative
Falsely
high
Folate
Quantitative
Falsely
high
aHAV IgM
Qualitative
Falsely
Negative
aHBc IgM
Qualitative
Falsely
Negative
Slide8Diagnostic Decision MakingThe medical specialty that nearly every practicing physician relies on every day, for which training in many medical schools is limited to no more than a scattered few lectures throughout the entire curriculum, is “laboratory medicine
.” Dr. Michael Laposata
Number and complexity of laboratory tests is large and growing as new tests become available in emerging subspecialties of laboratory medicine (e.g.,
pharmacogenomics
).Over a 90 day period, Mayo Medical Labs added 56 new tests to their Test Directory of ~3,200 total tests related to over 2,400 diseases, syndromes, conditions, and infectious agents.~90% of all laboratory tests (AP + CP) ordered are laboratory medicine (CP) testsThese tests are associated with 60-70% of all critical decision-making such as:Patient admittanceDischargeDrug therapyWithout sufficient knowledge of laboratory tests, health care providers are more prone to:Inappropriate test orderingMistakes in interpreting test results
(i.e., what do the results mean?)Poor case management
Increased costs per patientAdverse outcomes (Jennifer
Rufer
case
)
“America’s first clinical biochemist,” and among the fathers of the subspecialty of laboratory medicine known today as “clinical chemistry.”
Slide9How important are clinical laboratory test resultsin patient care outcomes?
Is there any other medical specialty with such a scope of influence on medical care in the U.S.?
From: Lewin Group: Laboratory Medicine: A National Status Report
Slide10Medical NecessityOrdering the minimum number of appropriate laboratory tests necessary to provide optimal patient care.
Reduce, if not eliminate, the ordering of “unnecessary” laboratory tests.What constitutes an “unnecessary” laboratory test? Any test for which, a priori, the results are not likely to be useful in the appropriate management of a patient’s medical condition.
Slide11Questions Clinicians Should Ask Before Ordering a Laboratory Test
Why is the test being ordered?What are the consequences of not ordering the test?How good is the test in discriminating between health vs disease?How are the test results interpreted?How will the test results influence patient management and outcome?
Source:
Rudolf RW et al.
Am J Clin Pathol. 2017;148:128-135.
Slide12“Legitimate” Reasons for Ordering a Laboratory Test
Diagnosis (to rule in or rule out a diagnosis)Monitoring (e.g., the effect of drug therapy)
Screening
(e.g., for congenital hypothyroidism)Research (understand pathophysiology of Dx)
Slide13Approaches to Establishing a Diagnosis Based on Laboratory Test Results
Hypothesis-deductionPattern recognition – a key component of Artificial Neural Networks (ANNs)Medical algorithmsRifle versus shotgun approach (to ordering laboratory tests)
Slide14Approaches to Establishing a Diagnosis Based on Laboratory Test Results
Hypothesis-Deduction (HD)PatternRecognition (PR)
Medical
Algorithms (MA)
Rifle vsShotgun (RS)Based on patient-specific non-laboratory findings, hypothesize what conditions (differential diagnoses) could cause these findings, followed by deducing the most likely cause based on the results of appropriate laboratory, and other types of, tests.Matching patients’ results for selected lab tests to results observed typically in a variety of disorders associated with the patient’s principal lab test finding (e.g., thrombocytopenia in a pregnant woman)Diagnosis of systemic infection (sepsis) and 28-day all-cause mortality risk based on procalcito-nin test results.Discriminately ordering only 7 lab tests known to have adequate diagnostic accuracy and predictive value in identifying a particular disease vs indiscriminate ordering of 20 lab tests, some or none of which may have clinical utility in identifying the disease.
Procalcitonin
Medical Algorithm
Slide15Another Example of Pattern Recognition: CSF Indices
The
number of possible permutations
of “A” and “N” for 4 indices is
24 or 16.
Slide16Clinical Performance Characteristics of Laboratory Tests
How do clinicians improve prevalence when ordering lab tests?
Slide17Receiver-Operator Characteristic (ROC) CurvesAUC values range from 0.5 to 1.0 and provide a quantitative representation of overall test accuracy:
0.5 to 0.7 (low accuracy) 0.7 to 0.9 (possibly useful) >0.9 (high accuracy)The ROC
curves below indicate that PSA (AUC = 0.66) has significantly higher diagnostic accuracy than PAP but only modest power in discriminating
BPH from organ-confined prostate
cancer:The sensitivity and specificity of a test is of limited value because these parameters represent the answer to the question:What is the probability of a patient having a positive test result if this patient has disease X?The more challenging question facing clinicians, however, is:What is the probability of this patient having disease X if the test result is positive (or negative)? Note: requires calculation of likelihood ratios.Diagnostic accuracy refers to the sensitivity AND specificity of a laboratory test.Area-under-the-(ROC) curve (AUC) values are a powerful tool in comparing diagnostic accuracy between 2 or more laboratory tests.
Diagonal
represents a test with no diagnostic value (AUC = 0.50)
PSA, prostate-specific antigen
PAP = prostatic acid
phoshpatase
Slide18Reference Interval for Interpreting Laboratory Test ResultsReference interval
(RI), more precisely, population-based, as opposed to individual-specific RI, refers to the two values representing the 2.5 and 97.5 percentile values for all values obtained on a sufficient number of “apparently healthy” individuals and these values are normally (aka Gaussian) distributed.
Stratified RI (e.g., based on sex and/or age)
Decision level
refers to a single cutoff value for a test that provides the highest discriminatory power in identifying samples from individuals with disease from those without the disease.
Slide19Critical Difference Between Consecutive Laboratory Test ResultsSources of variation affecting laboratory test results within and between individuals, and dates of testing include:
Analytical errorRandom error (RE)Systematic error (SE)Constant error (CE)
Proportional error (PE)
Biological (intra-individual) variation (CV
I) – an often forgotten source of variation in laboratory test results
Slide20Sources of Biological Variation (BV)
Chronological ageAlkaline phosphatase (ALP) values increase during pubertyGenderTestosterone values are higher in healthy males than in healthy femalesEstradiol values are higher in healthy, premenopausal females than in healthy males
Pulsatile and circadian biorhythms
Pulses of luteinizing hormone (LH) can be detected about every 90 min
ACTH and cortisol concentration begin to rise early in the morning, peak at ~8 am, and gradually fall over the course of the day, reaching a nadir at midnight Seasonal variationALT levels are ~12% higher in the Spring than in the FallGeographic variationCarboxyhemoglobin levels are higher in individuals living in areas with heavier automobile traffic than in individuals living in rural areas
Slide21What are the effects of BV on test results for individuals?Serially measured test results may be different at different times of the:
DayMonthSeasonGeographic locationSerially measured test results for some analytes
may remain relatively constant regardless of the day/month/season/geographic location
Sodium and potassium values vary very little in healthy individuals
Population-based reference intervals are not applicable to all individuals and all analytes regardless of the timing of the sample for testing
Slide22The Conundrum of Serial Testing
Physicians frequently order the same test at multiple time points (i.e., serial testing) during the course of the patients’ management.Laboratory test results for any analyte change over time within and between individuals (see graph below
):
Mean
Values and Ranges of Serum Creatinine Values in Four Samples at Different Time Points From 10 Healthy Men (dotted lines indicate reference interval: 0.72-1.36 mg/dL)The magnitude of the change is influenced by both analytical and biological sources of variation [i.e., analytical imprecision (CVA) and biological variability (CVI)]
Slide23The Conundrum of Serial TestingThe critical question facing clinicians are:
Is the magnitude of the change in serially measured laboratory test values on my patient statistically significant?If the probability that the change is statistically significant is high (e.g. > 80%), does this mean that some patient medical intervention is necessary, desirable, or strongly encouraged?
The answer to both of these questions requires an understanding of how to calculate the
critical difference
(CD), preferred term is now: reference change value (RCV), which is a presentation for another day, especially, given the current circumstance that the vast majority of clinicians are either not aware or are not enamored of the value of the RCV in their medical decision making.
Slide24Artificial Neural Networks (ANNs)
ANN*
A computer-derived system that uses multiple data inputs for pattern recognition of a clinical event (e.g., the likelihood of developing prostate cancer
).
*See Demirci F et al. Artificial neural network approach in laboratory reporting. Am J Clin Pathol. 2016;146:227-337. A superb article on how ANNs are created using multiple criteria for rejecting laboratory test results.
Slide25Final ThoughtsIn the final analysis, it is important for clinicians and laboratorians to recognize that laboratory data, although potentially extremely useful in diagnostic decision making, should be used as an aid and adjunct to the constellation of findings (
eg, history, physical exam, etc.) relevant to the patient.Data are never a substitute for a good physical exam and patient history (clinicians should treat the patient, not the laboratory results).
Slide26SummaryRather than a traditional Summary, the “Jennifer Rufer Case” illustrates with a real-world example the critical role medical laboratory individuals play daily in the practice of medicine and the treatment of patients, no less in accordance with the oath, “Do no harm,” taken and practiced by physicians.
Slide27The Jennifer Rufer Case
22 y.o., recently married, using contraceptives, presented with irregular bleeding between menstrual periodsSerum hCG concentration was elevatedUltrasound showed no fetal sacWorked up for ectopic
pregnancy
Slide28History of Case ContinuedLaparoscopy was negative for ectopic pregnancy
Dilatation and curettage (D&C) performed with histopathology indicating the presence of only normal menstrual tissueJuly 1998: [hCG] still elevated - trophoblastic choriocarcinoma suspected
Chemotherapy administered
with no effect on [
hCG] (i.e., remained elevated)Hysterectomy performed with no pathological abnormalities notedRare, resistant (to ChemoRx) form of choriocarcinoma suspected with possible metastasis to extrauterine locationAttending physician became suspicious about persistently elevated [hCG]Patient’s serum sample sent to 2 referral labs for hCG testingBoth referral labs used the same hCG immunoassay as the initial lab! hCG results elevated.
Slide29History of Case ContinuedMore chemotherapy administered
using 5 chemotherapeutic agentsCT scan performed which demonstrated a suspicious lung noduleLung lobectomy performed
Finally, the validity of the
hCG
results was questioned after a total of >45 hCG determinations had been performed on the patient’s serum during the course of her treatmentPatient’s persistently elevated hCG levels attributed to false-positive (FP) test results due to HAMA interference
Slide30Prevention, Detection, and Treatment to Reduce or Eliminate HAMA InterferencePrevention of HAMA production
Blocking HAMARemoving HAMAQuantifying HAMADiluting HAMA
Re-assay of patient’s sample with an immunoassay that uses different antibodies (e.g., goat antibodies) from the initial immunoassay
Redesign of immunoassays by manufacturers to eliminate HAMA interference
Treat patients with immunosuppressant Rx before, during, and after the administration of mouse antibody agentsCsACyclophosphamideAzathioprineDeoxyspergualin
Slide31History of Case Continued
SummaryPatient underwent:D&CHysterectomy
A round of initial
ChemRx
A second, more aggressive, round of ChemoRxLung lobectomyDue, in part, to FP hCG results and unrecognized human anti-mouse antibody (HAMA) interference
Presumably, there was no consultation with clinical pathology staff?
Why
?
A survey of 1,768 primary care physicians indicated*:
In 14.7% of patient encounters requiring laboratory testing there was uncertainty about which test(s) to order.
In 8.3% of these encounters there was uncertainty in interpreting the lab test results.
When physicians do not know
which lab test to order
or
what the results mean
, they consult a “lab professional” only 6% of the time, despite being satisfied with the outcome >50% of the time.
*
Hickner
J,
et al.
J
Am Board Fam Med
. 2014;27:268–274.
Slide32Thank You For Your Attention!
Behind every doctor is a great clinical laboratory scientist
!