TestTest Sensitivity%Sensitivity% Specificity%Specificity%
ANAANA 9999 8080
dsDNAdsDNA 7070 9595
ssDNAssDNA 8080 5050
HistoneHistone 30-8030-80 5050
NucleoproteinNucleoprotein 5858 5050
SmSm 2525 9999
RNPRNP 5050 87-9487-94
PCNAPCNA 55 9595
1
Which one of these test is the best for SLE Dx?
علی رغم داعیه »پزشکی مبتنی بر شواهد« در •بسیاری
از دانشکده های پزشکی، این امر آنگونه که باید مورد توجه و پذیرش قرار نگرفته است.
دانشجویان حساسیت و ویژگی را می آموزند و یاد •می گیرند که برای تشخیص چگونه هنر شرح حال
گیری و معاینه را با آزمونهای تشخیصی درهم بیامیزند ولی از تئوری احتماالت و به کارگیری آن
خبری نیست!
به همین جهت گاهی اوقات از ادبیاتی استفاده •می شود که مانع نتیجه گیری درست در تشخیص
می شود.مثال گفته می شود که اگر کسی فالن جواب را •
L آن بیماری خاص را دارد مگر آنکه داشت پس حتماخالفش ثابت شود.
مطابقت دارد؟ آیا Bayesianاین گفته چقدر با منطق توجهی به ماهیت آزمون شد؟ آیا توجهی به خود
بیمار یا جمعیت مشابه او شد؟ آیا این جمله به این معناست که احتمال پیش از آزمون خیلی باالست؟
یا این مورد:
• SnNout• in a highly Sensitive
test, a Negative test rules out the disease
• SpPin• in a highly Specific
test, a Positive test rules in the disease
چیست؟Bayesianمنطق
در مثال قبل برای گذاشتن تشخیص، فقط به •یک جنبه از آزمون توجه شده است: این تست
L آنقدر ویژگی باالیی دارد که اگر مثبت شود حتمابیمار آن تشخیص خاص را دارد.
می گوید که هر فرد بیمار قبل Bayesianمنطق •از انجام آزمون یک احتمال مشخصی برای
بیماری خاصی دارد که با یک عامل تعدیل کننده L نسبت درست نمایی یا یا Likelihood ratio)مثال
LR.این احتمال افزایش یا کاهش می یابد )
Dr. Shahram Yazdani
66
Medical Decision MakingMedical Decision Making
Refining ProbabilityRefining Probability Decision AnalysisDecision Analysis Treatment and Testing ThresholdsTreatment and Testing Thresholds Cost-Effectiveness AnalysisCost-Effectiveness Analysis
راههای تشخیص در پزشکی
- استدالل پاتوفیزیولوژیک1- شناخت الگوی بیماری2- استدالل احتماالتی3
برای مورد اخیر باید لیست تشخیص افتراقی •بیماری را بطور کامل مدنظر داشته باشیم و احتمال بیماری را )قبل از انجام هر آزمونی(
محاسبه کنیم.
احتمال قبل از آزمون چگونه محاسبه می شود؟•
- بر اساس تجربه شخصی )با تمام خطاهایی که 1دارد(
- اطالعات چاپ شده:2الف- استفاده از شیوع بیماری به عالوه عالئم و
شکایات خاص بالینیب- استفاده از قانونهای پیش بینی کلینیکی
تخمین احتمال پیش از آزمون
Dr. Shahram Yazdani
1010
Why we routinely order Why we routinely order diagnostic testsdiagnostic tests
It’s what we always do: It’s what we always do: TraditionTradition The hospital has to make money The hospital has to make money
somehowsomehow: : Economic GainEconomic Gain I just wanted to know the test resultI just wanted to know the test result: :
CuriosityCuriosity Dr. “X” told me to do itDr. “X” told me to do it: : Hierarchy Hierarchy
We have to learn how to perform We have to learn how to perform procedures somehow: procedures somehow: Practice Practice
برای چه آزمون می کنیم؟
رسم و سنتما همیشه آزمون می کنیم.•هدف اقتصادیباالخره بیمارستان خرج دارد!•کنجکاویفقط می خواهم جوابش را بدانم. •سلسله مراتبیبه ما اینطور گفته اند.•آموزش و تمرینباید یادبگیریم چطور آزمون کنیم.•
احتمال پیش از آزمون به احتمال پس از تغییر•
آزمون.
Dr. Shahram Yazdani
1212
The most important reason we The most important reason we order diagnostic tests is to order diagnostic tests is to Refine ProbabilityRefine Probability
Dr. Shahram Yazdani
1313
Probability:Probability: is a number between 0 is a number between 0 and 1 )or 0% and 100%( that and 1 )or 0% and 100%( that expresses the likelihood of expresses the likelihood of something happening or being truesomething happening or being true
Dr. Shahram Yazdani
1414
Refining ProbabilityRefining Probability
Modifying our estimate of the Modifying our estimate of the likelihood of a disease likelihood of a disease through the application of through the application of diagnostic testsdiagnostic tests
Dr. Shahram Yazdani
1616
How the result of diagnostic How the result of diagnostic tests change the likelihood of a tests change the likelihood of a
particular diagnosisparticular diagnosis
What we thought before + test What we thought before + test information =what we think afterinformation =what we think after
Dr. Shahram Yazdani
1717
What we thought before\What we thought before\afterafter
Pretest probability:Pretest probability: the probability the probability that a patient has the disease before that a patient has the disease before undergoing a testundergoing a test
Posttest probability:Posttest probability: the probability the probability that a patient has the disease ,given that a patient has the disease ,given the result of a testthe result of a test
Dr. Shahram Yazdani
1818
How the result of diagnostic How the result of diagnostic tests change the likelihood of a tests change the likelihood of a
particular diagnosisparticular diagnosis
What we thought before + test What we thought before + test information = what we think afterinformation = what we think after
Pretest probability + Likelihood Ratio Pretest probability + Likelihood Ratio = Posttest Probability= Posttest Probability
The effect of test results on The effect of test results on the probability of disease the probability of disease
Angina onexertion
Dr. Shahram Yazdani
0.0 0.5 1.0
PretestProbability
PosttestProbability
e.g. Stress ECG
Perform Test
Probability of Disease
The effect of test results on the The effect of test results on the probability of disease probability of disease
Dr. Shahram Yazdani
0.0 1.0
PretestProbability
Probability of Disease
Posttest Probabilityafter test 2
Perform Test 2
Posttest Probabilityafter test 1
Perform Test 1
Overview of the Diagnostic Overview of the Diagnostic ProcessProcess
11stst Stage: Stage: Initial judgment: intuition, implicitInitial judgment: intuition, implicit Prior / pretest probabilityPrior / pretest probability: Based on experience : Based on experience
& knowledge& knowledge 22ndnd Stage: Stage:
Diagnostic tests: Gather more informationDiagnostic tests: Gather more information 33rdrd Stage: Stage:
Update the initial probability estimateUpdate the initial probability estimate Posterior / posttest probabilityPosterior / posttest probability
Dr. Shahram Yazdani
Disease EstimateDisease Estimate
Disease prevalenceDisease prevalence InformationInformation
Pretest probabilityPretest probability Diagnostic test )TPR, TNR(Diagnostic test )TPR, TNR(
Bayes' theoremBayes' theorem
Posttest probabilityPosttest probability
Dr. Shahram Yazdani
Dr. Shahram Yazdani
2323
Test InformationTest Information
Sensitivity and SpecificitySensitivity and Specificity Positive and Negative Predictive Positive and Negative Predictive
ValuesValues Likelihood RatiosLikelihood Ratios
شاخصهای یک آزمون تشخیصی
Dr. Shahram Yazdani
2525
Sensitivity and specificitySensitivity and specificity SensitivitySensitivity:: the proportion of the proportion of
patients patients withwith the disease who have the disease who have a a positivepositive test result test result
SpecificitySpecificity:: the proportion of the proportion of patients patients withoutwithout the disease who the disease who have a have a negativenegative test result test result
Dr. Shahram Yazdani
2626
Information for a dichotomous Information for a dichotomous testtest
True positive True positive
AAFalse positiveFalse positive
BB
False negativeFalse negative
CCTrue negativeTrue negative
DD
Disease Present Absent
Positive
Negative
TestResult
Sensitivity = A / (A+C)
Specificity = D / (B+D)
Dr. Shahram Yazdani
2727
Information for a dichotomous Information for a dichotomous testtest
True positive True positive
A = 103A = 103False positiveFalse positive
B = 16B = 16
False negativeFalse negative
C = 12C = 12True negativeTrue negative
D = 211D = 211
Disease Present Absent
Positive
Negative
TestResult
Dr. Shahram Yazdani
2828
Information for a dichotomous Information for a dichotomous testtest
True positive True positive
A = 103A = 103False positiveFalse positive
B = 16B = 16
False negativeFalse negative
C = 12C = 12True negativeTrue negative
D = 211D = 211
Disease Present Absent
Positive
Negative
TestResult
Sensitivity=103/(103+12)=89%
Specificity=211/(16+211)=93%
Dr. Shahram Yazdani
2929
Sensitivity and specificitySensitivity and specificity Limitation:Limitation: we don’t know who has we don’t know who has
the disease before the test ! the disease before the test ! Otherwise we wouldn’t need to Otherwise we wouldn’t need to order the diagnostic test.order the diagnostic test.
محدودیت حساسیت و ویژگی
حساسیت در گروه بیماران و ویژگی در گروه •سالمها )غیر بیماران( تعریف می شود در
صورتیکه ما نمی دانیم فردی که به ما مراجعه کرده بیمار است یا نه؟!
ما با فردی برخورد داریم که جواب آزمایش او •مثبت یا منفی است. پس به ارزش اخباری
آزمونها بپردازیم ...
Dr. Shahram Yazdani
Predictive valuesPredictive values
PPV :PPV : the proportion of patients with a the proportion of patients with a positivepositive test result who test result who have have the the diseasedisease
NPV :NPV : the proportion of patients with a the proportion of patients with a negativenegative test result who do test result who do not havenot have the diseasethe disease
3131
Dr. Shahram Yazdani
Information for a dichotomous Information for a dichotomous testtest
True positive True positive
AAFalse positiveFalse positive
BB
False negativeFalse negative
CCTrue negativeTrue negative
DD
3232
Disease Present Absent
Positive
Negative
TestResult
Sensitivity = A / (A+C)
Specificity = D / (B+D)
PPV = A / (A+B)
NPV = D / (C+D)
Dr. Shahram Yazdani
Information for a dichotomous Information for a dichotomous testtest
True positive True positive
A = 103A = 103False positiveFalse positive
B = 16B = 16
False negativeFalse negative
C = 12C = 12True negativeTrue negative
D = 211D = 211
3333
Disease Present Absent
Positive
Negative
TestResult
Sensitivity=103/(103+12)=89%
Specificity=211/(16+211)=93%
PPV = 103 / (103+16) = 86%
NPV = 211 / (12+211) = 94%
Dr. Shahram Yazdani
Predictive valuesPredictive values
Limitation:Limitation: predictive values are predictive values are dependent on the fixed prevalence dependent on the fixed prevalence )pretest probability( of disease in the )pretest probability( of disease in the studied population.studied population.
If the pretest probability of the disease If the pretest probability of the disease is equal to prevalence of disease then is equal to prevalence of disease then the post test probability of disease will the post test probability of disease will be equal to PPV )e.g in screening(be equal to PPV )e.g in screening(
3434
ارزش اخباری آزمونها هم مفهوم حساسیت و •ویژگی و هم شیوع بیماری را در خود مستتر
دارد.در واقع اگر فرض کنیم که شیوع بیماری در •
جامعه همان احتمال پیش از آزمون باشد، ارزش اخباری مثبت، احتمال پس از آزمون
می شود.
محدودیت ارزش اخباری
ارزش اخباری به شیوع بیماری در جامعه •بستگی دارد در حالیکه شیوع بیماری تغییر
L ارزش اخباری برای آزمون می کند. مثال سال گذشته با امروز که 10تشخیص ایدز در
L فرق شیوع بیماری بیشتر شده است حتمامی کند.
پس سراغ شاخص نسبت درست نمایی برویم •که ...
Dr. Shahram Yazdani
Likelihood ratioLikelihood ratio Likelihood ratio = the likelihood of Likelihood ratio = the likelihood of
a test result in patients a test result in patients withwith the the disease / the likelihood of a test disease / the likelihood of a test result in people result in people withoutwithout the the diseasedisease
LR)+( = sensitivity/)1-specificity(LR)+( = sensitivity/)1-specificity( LR)-( = )1-sensitivity(/specificityLR)-( = )1-sensitivity(/specificity
3737
Dr. Shahram Yazdani
Information for a dichotomous Information for a dichotomous testtest
True positive True positive
AAFalse positiveFalse positive
BB
False negativeFalse negative
CCTrue negativeTrue negative
DD
3838
Disease Present Absent
Positive
Negative
TestResult
Sensitivity = A / (A+C)
Specificity = D / (B+D)
PPV = A / (A+B)
NPV = D / (C+D)
LR(+) = A /(A+C)
B / (B+D)
LR(-) = C /(A+C)
D / (B+D)
= sn / (1-sp)
= (1-sn) / sp
Dr. Shahram Yazdani
Information for a dichotomous Information for a dichotomous testtest
True positive True positive
A = 103A = 103False positiveFalse positive
B = 16B = 16
False negativeFalse negative
C = 12C = 12True negativeTrue negative
D = 211D = 211
3939
Disease Present Absent
Positive
Negative
TestResult
Sensitivity=103/(103+12)=89%
Specificity=211/(16+211)=93%
PPV = 103 / (103+16) = 86%
NPV = 211 / (12+211) = 94%
LR(+) = A /(A+C)
B / (B+D)
LR(-) = C /(A+C)
D / (B+D)
= sn / (1-sp)=12.7
= (1-sn) / sp=0.11
نسبت درستنمایی
شاخصی بین صفر و مثبت بینهایت••LR+ یعنی احتمال مثبت شدن آزمون در
بیماران تقسیم بر احتمال مثبت شدن آزمون در غیر بیماران
)یا حساسیت تقسیم بر یک منهای ویژگی(•LR- یعنی احتمال منفی شدن آزمون در
بیماران تقسیم بر احتمال منفی شدن آزمون در غیر بیماران
)یا یک منهای حساسیت تقسیم بر ویژگی(
Positive Likelihood Ratios
• It can also be written as thetrue positive rate/false positive rate.
• Thus, the higher the positive likelihood ratio, the better the test (a perfect test has a positive likelihood ratio equal to infinity).
Negative Likelihood Ratio
• It can also be written as thefalse negative rate/true negative rate.
• Therefore, the lower the negative likelihood ratio, the better the test (a perfect test has a negative likelihood ratio of zero).
(LR)نسبت درستنمایی
- مقدار این شاخص مستقل از شیوع بیماری 1در جمعیت تحت مطالعه ای است که این
شاخص در آن برآورد شده است. پس کافیست که شما احتمال پیش از آزمون بیمار
خود را بدانید.
2 -LR برای آزمونهایی که جواب آنها بیش از دو حال دارد نیز قابل محاسبه است. مثال برای
برای هر جواب اسکن LRتشخیص آمبولی ریه V/Q:به این ترتیب است
18.3برای احتمال باال –1.2برای احتمال متوسط –0.36برای احتمال پایین –0.1برای جواب طبیعی –
3 -LR های چند تست غیر وابسته را می توان در هم ضرب کرد و احتمال پس از انجام آزمونها
را حساب کرد.
محاسبه احتمال پس از آزمون
استفاده از نوموگرام•انجام محاسبات ریاضی•
(p)احتمال پیش از آزمون
(p/1-p)شانس پیش از آزمون
ضربدر نسبت درستنمایی(x LR1 x LR2 x … x LRn)
(odds)شانس پس از آزمون
(odds/1+ods)احتمال پس از آزمون
Bayes Theorem
Post-test Odds =
Likelihood Ratio X Pre-test Odds
48
Using Likelihood Ratios to Determine Post-Test Disease Probability
Pre-test probability of disease
Pre-test odds of disease
Likelihood ratio
Post-test odds of disease
Post-test probability of disease
Dr. Shahram Yazdani
Calculating posttest Calculating posttest probabilityprobability
4949
Pretest probability
Pretest odds Likelihood ratio Posttest odds
Posttest probability
=
Test Information
Odds=p/(1-p) P=odds/(odds+1)
Dr. Shahram Yazdani
Quantitative estimate of Quantitative estimate of posttest probabilityposttest probability
1.1. Estimate the pretest probabilityEstimate the pretest probability
2.2. Convert pretest probability to Convert pretest probability to pretest oddspretest odds
3.3. Multiply the pretest odds by the Multiply the pretest odds by the likelihood ratio to get posttest oddslikelihood ratio to get posttest odds
4.4. Convert the posttest odds to a Convert the posttest odds to a posttest probabilityposttest probability
5050
Dr. Shahram Yazdani
5151
0.1
0.2
0.5
1
2
5
10
20
304050
Pre-test probability
Post-testprobability
99
95
90
8070
60504030
20
10
5
2
1
0.5
0.2
0.1
6070
80
90
95
99Likelihood
ratio
100
5020
10521
0.50.20.10.050.02
0.01
0.005
0.0020.001
200500
1000
Nomogram for interpretingDiagnostic test result
مزایای نسبت درستنمایی
یادگیری و استفاده از آن آسان است.•در مورد آزمونهایی که بیش از دو جواب دارد نیز •
کاربرد دارد.مقدار عددی آن نیز کمک کننده است یعنی مثال •
باشد به احتمال زیاد جواب مثبت آن 10اگر باالی باشد، 0.1باعث تایید تشخیص و اگر کمتر از
جواب منفی اش رد کننده تشخیص است. ها(، LR برای چند آزمون)ضرب LRاستفاده از •
ساده تر از محاسبه حساسیت و ویژگی برای هر آزمون است.
Dr. Shahram Yazdani
Likelihood ratioLikelihood ratio
00
0.10.1
11
1010
++∞∞
Posttest probabilityPosttest probability
No diseaseNo disease
Lower Lower
UnchangedUnchanged
HigherHigher
Disease certainDisease certain
5353
A test with LR greater than 1, increases the probability of Disease; And a test with LR of less than 1, decreases the probability of disease.
•LR تغییر 0/1 و کوچکتر از 10 های بزرگتر ازاساسی در احتمال پیش از آزمون می دهند.
•LR تغییرات متوسطی در 0/2تا 0/1 و 10 تا 5 های احتمال پیش از آزمون می دهند.
•LR تغییرات کوچک )ولی 0/5تا 0/2 و 5 تا 2 هایگاهی با اهمیت( در احتمال پیش از آزمون می دهند.
•LR تا یک تغییرات کوچک )و غالبا 0/5 و 2 های یک تابی اهمیت( در احتمال پیش از آزمون بوجود می
آورند.
Values of Positive and Negative Likelihood Ratios )LR(
LR Poor-fair Good Excellent
Positive likelihood
ratio2.1-5 5.1-10 >10
Negative likelihood
ratio0.5-0.2 0.19-0.1 <0.1
Likelihood Ratios & You
• Allows us to determine the accuracy with which a test identifies the target disorder
• As the LR becomes larger, the likelihood of the target disease increases:Likelihood ratio Interpretation
>10 Strong evidence to rule in disease
5-10 Moderate evidence to rule in disease
2-5 Weak evidence to rule in disease
0.5-2 No significant change in the likelihood of disease
0.2-0.5 Weak evidence to rule out disease
0.1-0.2 Moderate evidence to rule out disease
<0.1 Strong evidence to rule out disease
TestTest SensitivitySensitivity SpecificitySpecificity LR(+)LR(+)
ANAANA 9999 8080 4.954.95
dsDNAdsDNA 7070 9595 1414
ssDNAssDNA 8080 5050 1.61.6
HistoneHistone 30-8030-80 5050 1.11.1
NucleoproteinNucleoprotein 5858 5050 1.161.16
SmSm 2525 9999 2525
RNPRNP 5050 87-9487-94 3.8-8.33.8-8.3
PCNAPCNA 55 9595 11
57
Which one of these test is the best for SLE Dx?
این را هم در نظر داشته باشید:کی برای بیمار انجام آزمون الزم
است؟
احتمال پیش از آزمون
صفر درصد درصد100
نه آزمون می خواهدنه درمان.
آزمون نمی خواهد،درمانش کنید.
آزمون کنید و بر اساس نتایجدر مورد درمان تصمیم بگیرید.
Clinical interpretation of post-test probability
59
Don't treat for disease
Do further diagnostic
testingTreat for disease
Probability of disease:
0 1
Testing threshold
Treatment threshold
Disease ruled out
Disease ruled in
If you are here, Test will help you to go toward one
end of this probability, either 0 or 1 to get the final decision.
فقط از آزمونی استفاده کنید که نتایج مثبت یا •منفی آن، احتمال پیش از آزمون را در اطراف
حد آستانه درمان جابجا کند.فرض کنید احتمال پیش از آزمون برای بیمار •
درصد باشد و حد آستانه درمان برای او 10 درصد باشد. پس آزمونی بایدانجام داد که 50
داشته باشد 5الاقل نسبت درستنمایی برابر وگرنه جواب مثبت آزمایش، احتمال وجود
درصد نخواهد رساند و کمکی 50بیماری را به به تصمیم گیری درمانی نمی کند.
Advantages of LRs The higher or lower the LR, the higher
or lower the post-test disease probability
Which test will result in the highest post-test probability in a given patient?
The test with the largest LR+ Which test will result in the lowest post-
test probability in a given patient? The test with the smallest LR-
62
Dr. Shahram Yazdani
6363
Managing diagnostic Managing diagnostic uncertaintyuncertainty
You are consulted to visit a 62-year-You are consulted to visit a 62-year-old man with 3 months history of old man with 3 months history of severe back pain. His weight severe back pain. His weight remained stable. CBC and routine remained stable. CBC and routine biochemistry were normal. ESR was biochemistry were normal. ESR was 52 mm / hour. An x-ray of the lumbar 52 mm / hour. An x-ray of the lumbar and thoracic spine was reported to and thoracic spine was reported to showing degenerative changes.showing degenerative changes.
what is your approach to this what is your approach to this patient?patient?
Dr. Shahram Yazdani
6464
Clinical findings predicting cancer as a cause Clinical findings predicting cancer as a cause of back pain.of back pain.
Finding Finding Age > 50 yearsAge > 50 years Unexplained weight lossUnexplained weight loss Previous history of cancerPrevious history of cancer Persistent pain despite 1 month of Persistent pain despite 1 month of
treatmenttreatment Duration of this episode > 1 monthDuration of this episode > 1 month Severe painSevere pain ESR > 20ESR > 20 ESR > 50ESR > 50 ESR > 100ESR > 100 Hematocrit < 30%Hematocrit < 30% Lytic or blastic lesion on spine x-rayLytic or blastic lesion on spine x-ray
LRLR 2.72.7 2.72.7 14.714.7 3.03.0 2.62.6 1.61.6 2.42.4 19.219.2 55.555.5 15.215.2 120120
Dr. Shahram Yazdani
6565
Clinical and Laboratory findings will change Clinical and Laboratory findings will change the disease probability just like a testthe disease probability just like a test
Given that the probability of Given that the probability of malignancy as the cause of malignancy as the cause of persistent back pain in the general persistent back pain in the general population is about 0.3%, what is population is about 0.3%, what is the effect of patient’s ESR on the the effect of patient’s ESR on the probability of malignancy in this probability of malignancy in this patient?patient?
Dr. Shahram Yazdani
6666
Clinical findings predicting cancer as a cause Clinical findings predicting cancer as a cause of back pain.of back pain.
Finding Finding Age > 50 yearsAge > 50 years Unexplained weight lossUnexplained weight loss Previous history of cancerPrevious history of cancer Persistent pain despite 1 month of Persistent pain despite 1 month of
treatmenttreatment Duration of this episode > 1 monthDuration of this episode > 1 month Severe painSevere pain ESR > 20ESR > 20 ESR > 50ESR > 50 ESR > 100ESR > 100 Hematocrit < 30%Hematocrit < 30% Lytic or blastic lesion on spine x-rayLytic or blastic lesion on spine x-ray
LRLR 2.72.7 2.72.7 14.714.7 3.03.0 2.62.6 1.61.6 2.42.4 19.219.2 55.555.5 15.215.2 120120
Dr. Shahram Yazdani
6767
Calculating posttest Calculating posttest probabilityprobability
Pret. p=0.003
Pret. Odds: 0.003 LR: 19.2 Postt. Odds: 0.0576
Postt p: 0.054
=
Odds=p/(1-p) P=odds/(odds+1)
Pretest odds×likelihood ratio=posttest odds
Dr. Shahram Yazdani
6868
Clinical and Laboratory findings will change Clinical and Laboratory findings will change the disease probability just like a testthe disease probability just like a test
Consider that x-ray of spine in this Consider that x-ray of spine in this patient shows a lytic lesion then patient shows a lytic lesion then what will be the probability of what will be the probability of malignancy in this patient malignancy in this patient considering also patients age and considering also patients age and ESR?ESR?
Dr. Shahram Yazdani
6969
Clinical findings predicting cancer as a cause Clinical findings predicting cancer as a cause of back pain.of back pain.
Finding Finding Age > 50 yearsAge > 50 years Unexplained weight lossUnexplained weight loss Previous history of cancerPrevious history of cancer Persistent pain despite 1 month of Persistent pain despite 1 month of
treatmenttreatment Duration of this episode > 1 monthDuration of this episode > 1 month Severe painSevere pain ESR > 20ESR > 20 ESR > 50ESR > 50 ESR > 100ESR > 100 Hematocrit < 30%Hematocrit < 30% Lytic or blastic lesion on spine x-rayLytic or blastic lesion on spine x-ray
LRLR 2.72.7 2.72.7 14.714.7 3.03.0 2.62.6 1.61.6 2.42.4 19.219.2 55.555.5 15.215.2 120120
Dr. Shahram Yazdani
7070
Calculating posttest Calculating posttest probabilityprobability
Pret. p=0.003
Pret. Odds: 0.003 LR: 2.7×19.2×120 Postt. Odds: 18.6
Postt p: 0.94
=
Odds=p/(1-p) P=odds/(odds+1)
Pretest odds × LR1 × LR2 × LR3=posttest odds
Dr. Shahram Yazdani
7171
Predictive ValuesPredictive Values
Alternate formulations: Alternate formulations: Bayes’ TheoremBayes’ Theorem
PV+ =PV+ =
Se Se Pre-test Prevalence Pre-test Prevalence
Se Se Pre-test Prevalence + )1 - Pre-test Prevalence + )1 - SpSp( ( )1 - Pre-test )1 - Pre-test Prevalence(Prevalence(
High specificity to “rule-in” diseaseHigh specificity to “rule-in” disease
PV- =PV- =
Sp Sp )1 - Pre-test Prevalence( )1 - Pre-test Prevalence(
Sp Sp )1 - Pre-test Prevalence( + )1 - )1 - Pre-test Prevalence( + )1 - SeSe( ( Pre-test Pre-test PrevalencePrevalence
High sensitivity to “rule-out” diseaseHigh sensitivity to “rule-out” disease