Diagnostic tests

1
Diagnostic tests

• Charlotte Glümer
• Bendix Carstensen
• STAR-course
• Epidemiology and Biostatistics
• Mumbai, February 2003

2
Evaluation of a test

• Ideal world

• Real world

Test positive sick
Test negative healthy
Test negative healthy
Test positive sick
3
Test performance(1)
Sensitivity a/ac, Specificity d/bd Pv
a/ab, Pv d/cd
4
Test performance (2)

• Sensitivity How many of the diseased will be
  detected by the test.
• Specificity How many of the non-diseased will be
  classified as normal.
• Positive predictive value (PV) If the test is
  positive what is then the probability of being
  sick
• Negative predictive value (PV-) if the test is
  negative what is then the probability of being
  healthy

5
Sensitivity vs. specificity
6
Sensitivity vs. specificity
7
Reference test/Gold standard ?

• A diagnostic procedure that classifies
  individuals in the true categories diseased or
  non-diseased.
• Cancer histological verification
• DM Blood glucose
• Fasting plasma glucose
• 2 hour plasma glucose

8
Diabetes Diagnosis
9
FPG and RetinopathyNHANES (USA)
ADA, Expert Committee, Diabetes care 1997
10
Reclassification and mortality
Hazard Ratio
FPG (mmol/l)
2hPG (mmol/l)
DECODE-study
11
Misclassification
DECODE-study, 1998
12
Diabetes diagnoses

• WHO (1999)
• FPG ? 7.0 mmol/l
• OR
• 2hPG ? 11.1 mmol/l (200mg/dl)

• ADA (1997)
• FPG ? 7.0 mmol/l (126mg/dl)

Epidemiological diagnoses One value exceeding
threshold Clinical diagnoses Symptoms one
diabetic value OR two diabetic values on
two separate days
13
Advantages/disadvantages

• WHO
• Classifies all diabetics
• -
• Time consuming

• ADA
• Fast, one measurement
• -
• Misclassification of 1/3 of the diabetics

14
Evaluation of tests

• Gold standard Oral Glucose Tolerance Test (OGTT)
• Cut off value WHO (1999)
• FPG ? 7.0mmol/l or 2h-PG ? 11.1 mmol/l

15
Discussion Point

• What is sensitivity, specificity, and predictive
  values?
• What is the gold standard for diabetes ?
• How is it to be used?

16
Introduction to group work
17
Inter99

• Population based intervention study
• designed to answer
• Intervention part
• reduce the incidence of CVD and DM in a
  population at high risk.
• Screening part
• Develop and test screening strategies for Type 2
  diabetes

18
Inter99
All inhabitants aged 30, 35, 40, 45, 50, 55, and
60 years living in the Western Part of Copenhagen
County 61.301 people
Invitees
Non-Invitees
19
Inter99

• Participants
• 6784(52.5) participated
• mean age 46 years
• 51.3 females
• Screening
• Questionnaire
• Physical examination
• Blood tests incl. an OGTT