Transcript Journal Club - NCC Pediatrics Residency @ Walter Reed Bethesda

Journal Club
How to Use and Article About a
Diagnostic Test
Satyen Gada, MD
LT MC (FS) USN
NCC Pediatric Residency Program
Patients
Patient 1: 11 yo Caucasian male seen in
continuity clinic as a follow up from ER for CC of
H. pylori infection. Review of the ER sheet
significant for hx of burping with increased
frequency and halitosis over the last two days.
Family hx of uncle living in California dx with
gastritis 6 months ago. He was positive for H.
pylori on a blood test. Pt was started on therapy
for presumptive H. pylori gastritis. CBC, Stool
Guaiac, and H Pylori blood test sent.
Patients
Patient 1: In the office, hx from ER is
consistent. However, his snxs have
completely resolved. All further hx, ROS,
PE unremarkable. Lab results neg/wnl.
Mother concerned that the test was
performed incorrectly or not a good one.
According to the internet, her son must have
H. pylori, and he did get better on abx.
Patients
Patient 1: Your preceptor, a GI specialist,
tells you to write the note as she discusses
the case with the mother. After some time,
the mother storms out angry and upset.
Your preceptor shakes his head and throws
away two bottles of antibiotics.
Patients
Patient 2: The same day in clinic, you evaluate
another 11 yo male with abdominal pain. Pt has a
two week hx of worsening epigastric pain with
occasional emesis. Pt has had a three pound wt
loss. States his stool are now black. Increased
consumption of milk because it “feels better”.
Mother states that he was evaluated in the ER for
similar sxns 2 months ago and resolved with four
weeks of tx with Zantac. H. pylori blood test at
that time was negative. Mother states she also had
similar sxns and was diagnosed with an ulcer 1 yr
ago and given an antacid which seemed to help.
Patients
Patient 2: He is currently not taking any
medication.
Vital signs are stable, physical exam significant
for epigastric tenderness on deep palpation and
stool positive for occult blood.
Your preceptor, a GI specialist, tells you he/she
will take care of it and to go ahead and see the
triplet NICU f/u’s that have been waiting for 1 hr
while you obtained the above detailed hx.
Patients
Patient 2: After your NICU f/u, your preceptor
tells you despite the negative H pylori blood test,
this patient warrants further testing for H pylori
with a “breath test” in the GI clinic. You did not
appreciate the patient having bad breath, but agree
with the plan. When you put the order in AHLTA,
you notice there are four available tests for H
pylori: blood, urine, stool, and breath test.
Patients
Patient 2: The patient returns following the
GI appointment with a diagnosis of H pylori
peptic ulcer disease, and is being treated
with the same regimen that was given to
Patient #1 by the ER. The patient returns
for a school physical 10 months later and
has been sxn free.
Clinical Questions
Clinical Questions
1. Both patients had neg H. pylori blood
tests. Why was one patient referred to
more testing while the other told to
discontinue therapy?
2. Why was the breath test ordered for
patient #2 over the urine or stool?
3. How do I approach H. pylori testing
with future patients?
Hmm…
You return home a bit confused. Looking in
the mirror, your reminded of how MOTO
you are. You search MD consult to learn
more about H. pylori and H. pylori testing
and find an article titled….
Hmm…
Comparison of non-invasive tests to detect
Helicobacter pylori infection in children and
adolescents: Results of a multicenter
European study. Megraud et al. Journal of
Pediatrics Vol 146:2. Feb 2005.
Hmm…
While you are deleting your junk e-mail that
evening, you briefly read your Chief
Resident’s messages. One is about some
EBM articles on the nccpeds.com website.
You print out the section on how to use an
article about a diagnostic test.
H. pylori
 H. Pylori is a gram-negative spiral bacteria and is
estimated to infect more than half of the world’s
population, predominantly in developing
countries.
 Organism is well established as the cause of
gastritis-associated GI diseases: gastric ulcer,
duodenal ulcer, gastric cancer, and MALToma.
H. pylori
 H. pylori is one of the most common chronic
bacterial infections among humans.
 Most people are infected before age 10.
 Mode of transmission is person-to-person by
fecal-oral or gastro-oral routes.
 Primary caregivers are the reservoir for spread of
infection to children.
 In areas of poor sanitation, contaminated water or
food may be the primary reservoir.
H. pylori
 More prevalent in developing countries.
 Inverse relationship of prevalence with
socioeconomic status.
 Estimated prevalence in the US: African
Americans: 50%, Mexican-Americans
60%,Whites 26%
 Prevalence in Western Europe similar to
US.
H. pylori
 Variable latent period of sub-clinical infection
which causes gastric mucosal inflammation and
progressive mucosal damage, leading
symptomatic infection.
 Age of onset depends on virulence of strain and
age of initial infection.
 Lifetime risk for H. pylori-infected individual to
develop peptic ulcer disease is 1 in 6
 Risk for developing gastric cancer in countries
with high prevalence of H. pylori ranges from
almost 20% in China to 1.5% in the UK.
H. pylori
 Studies have shown that H. pylori does not
cause GERD, and eradication therapy for H.
pylori does not treat GERD long term.
 Those who are positive with H. pylori will
have recurrence of sxns unless H. pylori is
eradicated.
Diagnostic Tests for H pylori
infection
Tests are divided into those which require
endscopy (invasive) and those which do not
(non-invasive).
H. pylori
Invasive tests:
1. Gastric biopsies for H. pylori culture- gold
standard. (can also be obtained with gastric
brushings).
2. Biopsies for histology: evaluate severity of
gastritis and density of organism.
3. Urease testing: Add solution with urea: enzyme
converts urea to ammonia and change noticed on
pH indicator.
H. Pylori
1.
2.
3.
4.
Noninvasive tests:
Serologic testing: IgG ant-H. pylori antibodies
in serum-genrally present 4wks following
infection.
Stool antigen detection
Antibody detection in urine
Urea breath test: Drink solution containing urea
with labeled carbon. Urease activity by H.
pylori yields NH3 and CO2, which can be
measured with spectrometry.
H. pylori
Treatment: H2 blockers/PPI and:
Amoxicillin/Metronidazole +
Clarithromycin x 14d
Eradication rates are 60-80%
Tx failures should be re-treated with
Amoxicillin and PPI/H2 blockers + another
antibiotic.
Study: Methods



Open, prospective, multi-center study.
Inclusion criteria: Patients between ages 2 and
17 included if required an upper endoscopy and
H. pylori testing.
Exclusion criteria: Previous H. pylori
eradication tx, consumption of abx, antisecretory drugs, bismuth salts in last 2 weeks,
contraindication to endoscopy/biopsy.
Methods
 Estimated 600 children (2-11) and 200 adolescents
(12-17) needed to have at least 30 H. pylori
positive patients per age group.
 Study stopped when more than 30 H. pylori
positive patients would be included in each age
group.
 Within one week, endoscopic exam with biopsies,
13C urea breath test, stool antigen test, serology,
and antibody detection in urine were performed.
Methods
Definition of H. pylori status:
 Positive H. pylori defined as positive
culture from endoscopy specimens.
 In case of negative culture, positive results
of both histology and rapid urease test.
 Negative H. pylori if all invasive tests gave
negative results.
 Discrepant results were excluded.
Methods





Noninvasive testing
13C urea breath test
Stool antigen test
Serology
Antibody detection in urine
Methods: Are the results valid?
 Can you accept the reference standard? Yes:
Culture is gold standard for identification of H.
pylori. Studies have shown histology/urease
testing also to have high sensitivity (90-95%) and
specificity (95-100%)
 Where the reference standard and tests assess
independently of each other?
Yes: All patients received all tests and they were
performed blinded.
Methods: Are the results valid?
 Did the results of the test influence the
decision to perform the reference standard?
No: all tests were performed and blinded.
 Were the methods of performing the test
described in sufficient detail to permit
replication?
Yes: Name of test, company, and method of
storage/performance were outlined.
Methods: Are the results valid?
 Did the patient sample include an appropriate
spectrum of patients to whom the diagnostic test
will be applied in a clinical practice?
(see Table 1)
Pros
+ Age inclusion was broad 2-17. Good exclusion
criteria.
+ Good male/female ratio
+ Varied reasons for workup/findings on endoscopy
Methods: Are the results valid?
Cons
- Almost exclusively Caucasian.
- Small number in 2-5 age range
- Author comments on number of immigrant
children in test as well as the fact that all
patients warranted endoscopy.
- Patients from Western Europe: however,
prevalence similar to US.
Interpreting Results of Diagnostic
Testing
• Pretest probability: Start with the patient
presenting with the constellation of sxns and signs.
Even though two patients may have the same
results, the probability of those results meaning
anything is different-you may treat one and order
further studies in another.
• Pretest probability, after results are obtained, alters
post-test probability. Altered by the likelihood
ratio of the test.
Interpreting Results of Diagnostic
Testing
 Likelihood ratio (LR): How likely is it that the
given diagnostic test will raise or lower the post
test probability of the target disorder. LR of 1
means that the post test probability is exactly the
same as the pretest probability.
 Ratio of the probability of a positive test
confirming the disorder to a negative test reducing
the chance of a disorder.
 1/LR = unlikelihood ratio
Interpreting Results of Diagnostic
Testing
 LR greater than 10 or less than 0.1 generate
large and often conclusive changes from
pretest to post-test probability.
 5-10 or 0.1 to 0.2 generate moderate shifts.
 2-5 and 0.5 to 0.2 small changes.
 1-2 and 0.5 to 1 alter to small degree (rarely
important).
Interpreting Results of Diagnostic
Testing
 Normogram allows us to go from pretest to
post test probability.
 Pretest, LR, and Post Test Probability.
 Can use sensitivity and specificity to create
a receiver operator curve.
 Post test probability leads to
interpretation/treatement (or lack of).
What are the results?


Sensitivities, specificities, predictive
values and diagnostic accuracy were
determined in separately for each
diagnostic test.
ROC were also created
Results
Test is
positive
Test is
negative
Patients with Patients
disease
without
disease
a
b
c
d
Sensitivity= a/a+c
Specificity= d/b+d
Results
Positive/Negative predictive values depend
on prevelance of disease:
Low prevalence will yield more false
positives, and therefore lower positive
predictive value.
Low prevalence will yield more true
negatives, and therefore higher negative
predictive value.
Results
 503 recruited for study
 316 had results of all non-invasive tests and
fulfilled definition of +/- H. pylori status
 Of the 316, 42% positive for H. pylori.
 (High prevelance of disease)
Table II . Performances of the diagnostic tests for the 316 patients with gold standard and
four tests performed (UBT, HpSA, Urinelisa and Pyloriset EIA-G)
Age group (y)
Sensiti
vity
Specifi
city
Global
Global
Helicobacter test INFAI
HpSA
Urinelisa
Pyloriset EIA-G
Urea Breath
Stool
Urine
Blood
Urine 2
72.9 [64.980.0]
63.2 [54.771.0]
88.7 [82.593.3]
30.2 [22.538.9]
∗∗∗
80.3
[73.0-86.5]
∗∗∗
72.2
[64.1-79.3]
∗∗∗
97.3 [94.099.0]
97.3 [94.099.0]
93.4 [89.196.4]
∗∗∗
93.4
[89.1-96.4]
∗∗∗
93.4
[89.1-96.4]
∗∗∗
96.2 [91.9-98.6]
97.3 [94.0-99.0]
∗∗
Rapirun
90.2
[84.2-94.4]
98.7 [95.799.8]
93.9
[89.1-96.4]
Accura
cy
Global
96.8 [94.4-98.4]
87.0 [83.090.4]
82.9 [78.486.8]
91.5 [88.094.2]
68.7 [63.074.0]
PPV
Global
96.2 [91.9-98.6]
95.1 [89.598.2]
94.4 [88.097.9]
90.8 [84.894.9]
94.7 [83.699.1]
NPV
Global
97.3 [94.0-99.0]
83.2 [77.787.7]
78.4 [72.783.4]
91.9 [87.395.2]
64.5 [58.270.4]
Authors were aware of the high prevalence of H.
pylori. Conducted subgroup analysis of 76 patients
in low prevalence centers for PPV/NPV
Test
PPV (%)
NPV (%)
Urea Breath 76.4
98.3
Serum
76.4
98.3
Stool
83.3
93.7
Urine
83.3
93.7
Urine2
84.7
75
Results
ROC were created. Authors noted that
manufacturer threshold for postive/negative
results were not ideal.
Sensitivity for stool could be increased from
72.9 to 80.3 and urine from 63.2 to 72.2
with minimal effect on specificity.
Are the results applicable to my
patient?
Is the reproducibility high?
YES: Standard criteria for interpretation of
biopsies and manufacturer directions for
testing were used.
Did comment that degradation of samples in
shipment may have occurred when
transporting to sites (Germany, Belgium,
France).
Are the results applicable to my
patient?
 If you meet inclusion criteria and no
exclusion criteria, results are usually
applicable. Both patient 1&2 satisfy.
 Due to varied military population,
prevalence is difficult to obtain, therefore
affecting PPV and NPV.
Patients
The choice of tests depends upon issues
such as cost, availability, clinical situation,
prevalence of infection, pretest probability
of infection, and presence of factors (PPI
and abx) that may influence test results.
Patient 1&2
• There is a threshold in which a physician
will accept a test, ordering no further tests.
• When the probability of a target disorder
lies between the test and treatment
thresholds, further testing is mandated.
• Once we decide what our test and treatment
thresholds are, post test probabilities have
direct implications.
Patient 1
 No clear diagnosis-sxns more consistent with
transient dyspepsia.
 Uncle lives in California
 Lower prevalence in those with his sxns
 Pretest probability low
 Likelihood ratio/unlikelihood ratio low
 H. pylori serum AB with sensitivity of 88.7 and
specificity of 93.4% and NPV of 98.3% in group
with low prevalence.
 Overall post test probability low.
Patient 2






Pt with sxns and hx of recurrent ulcer.
Melena and emesis on hx.
Blood in stool confirmed on exam.
Suggestive family history of ulcer disease.
Likely high prevalence in this patient’s family.
Neg serum AB, sensitivity of 88.7%, PPV of 90.8
%,NPV of 91%.
 Urea breath test with sensitivity of 96.2%, PPV of
96.2% and NPV of 97.3% in high prevalence
group
 Better test in this patient and warranted.