Transcript Laboratory Test Utilization

Laboratory Test Utilization: The
Good, the Bad and the Overused
Tim Hamill, MD
Director, UCSF Clinical Laboratories
The Good, The Bad & The Ugly
• A ‘Good’ test:
• Provides information that is useful in patient
management decisions
• Screening: High sensitivity & NPV
• Diagnosis: High specificity & PPV
• A ‘Bad’ test:
• Uses resources but fails to provide information
useful in patient management decisions
• The ‘Ugly’ test:
• Uses resources and provides information that is
misleading or irrelevant
UCSF Test Utilization
• In 2009 UCSF performed an average of
486,000 tests per month
• Annual test volume 5.8M
• Inpatient tests : Outpatient tests = 1:1
• Inpatients: 15 tests/day
• Outpatients: 4 tests per visit
• Stat : Routine = 1:3 overall (1:1 inpatient)
UHC Comparison Data
UCSF Ranks #1 in Tests Used per Patient
Discharged
Causes of Test Overutilization*
• Ordering test panels rather than ala carte’
• Ordering tests as groups
• Repetitive test orders (esp. normal results)
• Incomplete understanding re: impact of low pre-test
probability
• Poor understanding of the consequences of overutilization
• Patient pressure
• Defensive testing
• Perverse financial incentives (more tests = more revenue)
* Astion ML. 2006. Interventions that improve laboratory utilization: from gentle guidance to strong
restrictions. Laboratory Errors and Patient Safety. 2(4):8-9
The Origin of Test Panels
• Technicon SMAC (1974):
• Sequential Multiple Analyzer with
Computer
Modern ‘Discreet’ Analyzers
Ordering Test Panels
• A study of orders for repeat electrolyte panels indicated that 10% were
medically unnecessary and in 65% of cases a single test could have
substituted for the entire panel. (Baigelman et al, Intensive Care Med, 11(6) 1985)
Comparison of Electrolyte Orders: Individual vs. Panel Orders
140000
120000
Test Volume
100000
80000
Overall total
Total as separate orders
60000
40000
20000
5013
4061
995
966
Na
K
Cl
CO2
0
Impact of Electrolyte Reduction
Ordering Tests in Groups
• Redundant tests:
• BUN & Creatinine
• Troponin & CK-MB
• ALT & AST
• Tests that just seem to trip off the tongue:
• Calcium, magnesium, phosphorus
• PT & PTT
• T3, T4 & TSH
Separating BUN & Creatinine
Repetitive Testing
• A study of the impact on serum potassium
orders using a simple algorithm based on
prior tests being normal or abnormal could
reduce potassium testing by 34% (Schubart et al,
MEDINFO 2001)
• Renal function: BUN, Creatinine
• CBC & CBC w/differential
Problems with Test Overutilization
• Patient issues
• Pain & morbidity from repeated venipunctures
• Iatrogenic anemia
• Medical issues
• Follow-up on clinically irrelevant abnormals
• Tracking just the ‘numbers’ instead of the entire
clinical picture
• Instituting inappropriate therapies
• Economic & Environmental issues
• Lack of reimbursement for inpatient testing
• Biohazardous waste generation
Studies on iatrogenic anemia
• Smoller et al NEJM 314, 1986:
• General wards: 1.1 draws/d, Ave. 12.4 mL/d, Total
175 mL for hospitalization
• ICU: 3.4 draws/d, Ave. 41.5 mL/d, Total 762.2mL
• ICU w/Art line: 4.0 draws/d, Total 944 mL
• Low et al Chest 108(1) Jul, 1995:
• Presence of Art. Line in ICU patients increased
blood volume loss from phlebotomy by 44%
Impacts of Iatrogenic anemia
• Critically ill patients may not have the bone
marrow reserve or erythropoietin drive to
compensate for iatrogenic blood loss.
• Transfusion to correct for this anemia has been
shown to negatively impact long term survival
• Other risks of phlebotomy:
• Nerve damage, arterial damage, venous sclerosis,
infection
Irrelevant ‘abnormals’
• Virtually all quantitative laboratory test ‘normal ranges’ are based on
the mean +/- 2 SD (95% confidence interval) for a subject population
2.5%
2.5%
• 5% of normal patients will have values that lie
outside this range (magnified for ill patients)
Irrelevant ‘abnormals’
• If a patient has 10 tests ordered, each with a
5% chance that the test may have a result
outside the normal range. Then there is a
50% chance that at least one test will have
an ‘abnormal’ result
• This is especially true with ordering
chemistry ‘panels’
Economics & Environment
• The vast majority of inpatient care is covered by
DRG or per diem payments
• Laboratory tests are not individually reimbursed and
merely represent cost against the what the hospital is
paid
• The 3 UCSF Clinical Laboratories generate
approx. 11,500 lbs of biowaste per month
• Cost to incinerate this waste is approx. $88K per year
The Solution?
• Approaches that have been tried
•
•
•
•
•
•
Place limits on housestaff orders
Provide information on test costs
Requisition design
EMR warnings and reminders
Education
Incentives
Factors that Impact Laboratory
Test Results & Interpretation
• Pre-analytic issues
• Diagnostic testing issues
•
•
•
•
•
Pre-test probability
Appropriateness of test in your patient
Impact of the test result on care decisions
Chasing ‘diagnostic certainty’
Impact of other disorders, therapy on results
• Monitoring issues
• Which test is going to be used?
• How fast does the test change?
• What is the impact of monitoring on clinical care?
Factors Impacting Laboratory
Tests
Pre-analytic Test Issues
•
•
•
•
Time of sample collection
Proper collection technique
Proper labeling
Proper storage and transport
Time of Collection
• Relative to time of management decisions
• Relative to therapy
• Platelet counts after transfusion
• Drug levels: relative to dose
• Peak: PO vs. IV
• Trough
• Relative to time of day
• Cortisol
Proper Collection Technique
• Prolonged use of tourniquet
• IV’s and line draws
• Order of collection
• Trace metal (Royal blue), Blood cultures
• Blue; Gold/Red; Green; Purple
• Proper filling
• Proper mixing
• Special needs
Proper Labeling
• The person who collects a sample should
label it
• Double check the label before submitting
• Special Blood Bank requirements
• Check specimen
• Body fluids
• Lab policies on handling unlabeled and
mislabeled samples
Proper Storage & Transport
• Refrigeration vs. Room temperature
• Protection from light
• Transport to the lab
• Effects of cellular metabolism
• Blood gas samples
• Serum chemistries
• Pneumatic tube considerations
What is the effect of Pre-test
probability?
• The pretest probability of disease is critically
important to test ordering
• If the disorder has a low pretest probability then even a
very sensitive and specific test may have little clinical
utility
• If the pretest probability is very high tests may not
provide much (if any) additional ‘certainty’ of the
diagnosis
Low pretest probability example
• You are thinking of ordering a test that is 95%
sensitive and specific
• The pretest probability that the disorder is present
is, however, only 10% in your patient
• Positive predictive value: 68%
• Negative predictive value: 99%
• only represents a 9% increase in certainty over the pretest
probability
• At a pre-test probability of 1% the PPV is only
16% and the NPV is 100%
High pretest probability
• The same theoretic test (95% sensitive and
95% specific) but in a patient with a 90%
likelihood of having the disorder:
• Positive predictive value: 99% (9% incr.)
• Negative predictive value: 68%
When do tests add the most
information?
• When the pretest probability is in the range
of 30-70%
• At a pretest probability of 50%, a test that is
95% sensitive and specific yields:
• Positive predictive value: 95%
• Negative predictive value: 95%
Is the test appropriate for your
patient?
• How does the patient’s diagnosis impact the test?
• VTE and Protein C/S levels
• How does the patient’s treatment impact the test?
• Galactose or Maltose containing medications and
POCT Glucose testing
• D-dimer in a surgical patient
• How do other disorders impact the test?
• HgbA1c and shortened red cell lifespan
Impact of the test on patient care
decisions?
• Ask yourself….’What will I do if the test is
negative/normal vs. positive/abnormal’ if
the answer is essentially the same for both
then the test has little utility
• E.g. Haptoglobin in anemia
• Will the result be available before the
patient is discharged?
• Chasing diagnostic ‘certainty’
• How much information is needed before a
treatment is initiated or withheld?
Questions to ask about tests used
to monitor a patient
• How fast do I expect the test to change?
• What is the best monitoring test for the disorder in
question?
• Is more than one test needed?
• How much change would trigger a therapeutic
intervention?
• Once an intervention is made is monitoring still
needed? With the same test? At the same
frequency?
How fast do tests change?
• While many test results can change rapidly in an
individual there are other tests which may only
change slowly or not at all over time
• Positive serologic studies
• Enzyme levels: AST T1/2 = 17 hr, ALT T1/2 = 48 hr, Alk
Phos T1/2 = 7 d,, GGT T1/2 = 9 d (28 d in hepatic Dz)
• D-dimers
• WBC differential (if no change in WBC)
• Understanding how rapidly a given analyte may
change is important to selecting how often it
should be ordered
Laboratory Manual
• The UCSF Clinical Laboratories maintain an online laboratory manual that is constantly updated
• http://labmed.ucsf.edu/labman/
• It has important information about:
• Sample type and amount, incl. minimums
• Test availability and turnaround time
• Patient preparation, collection instructions and sample
handling
• Test utilization tips