Examqo › Examqo library › Public health › Epidemiological methods · IMLE
Validity and reliability of diagnostic tests
Sensitivity and specificity belong to the test itself, whereas the predictive values depend on the prevalence of disease in the tested population – so in a rare disease the same test gives a very low positive predictive value.
Why it matters in the exam
This calculation topic appears in almost every exam: a two-by-two table, or sensitivity, specificity and prevalence, is given, and you must compute the positive and negative predictive values, a likelihood ratio, or what happens when the cut-off changes. Other items deal with serial testing (a second test only for positives), which raises specificity and the positive predictive value, and parallel testing, which raises sensitivity. Also tested are reliability versus validity, observed percent agreement versus the kappa statistic, which corrects for chance agreement, and the ROC curve. An example worth remembering: at a prevalence of 1%, a sensitivity and specificity of 90% give a positive predictive value of only about 8% – most positives are healthy. This explains why a screening test in the general population produces many false positives. Further items present a new test evaluated in hospital, in very sick patients versus clearly healthy people, and ask what will happen to sensitivity and specificity when it is used in the community. The answer is that both usually fall, because the spectrum of patients and healthy people in the community is less extreme – this is spectrum bias. A test validated against an imperfect gold standard may also look better or worse than it really is, so it matters what it was compared with.
Key points
- Sensitivity and specificity are test properties and do not depend on prevalence.
- The positive predictive value rises with prevalence and is therefore low in general-population screening.
- Lowering the cut-off raises sensitivity at the cost of specificity.
- Serial testing raises specificity; parallel testing raises sensitivity.
- Kappa measures inter-rater reliability beyond chance agreement.
Summary table
| Measure | Definition | Depends on prevalence? | Main use |
|---|---|---|---|
| Sensitivity | Share testing positive among the diseased | No | A sensitive test that is negative rules disease out |
| Specificity | Share testing negative among the healthy | No | A specific test that is positive rules disease in |
| Positive predictive value | Share diseased among those testing positive | Yes – rises with prevalence | The chance that a positive person is truly ill |
| Negative predictive value | Share healthy among those testing negative | Yes – falls as prevalence rises | Reassuring a person with a negative result |
| Positive likelihood ratio | Sensitivity divided by (one minus specificity) | No | Updating the pre-test probability |
| Kappa | Agreement between raters beyond chance | No | Inter-rater reliability |
A free exam question
בן 70 מתלונן על כאב ראש טמפורלי משמאל, חולשה, חום C 37.50 , כאב ב חגורת הכתפיים וללא תלונות עיניות. בבדיקתו: רגישות ניכרת במישוש עורק טמפורלי משמאל, בדיקת עיניים תקינה. במעבדה: אנמיה נורמוציטית 12 ג'/ד"ל)בסיס 14 ( , שקיעת דם 100 מ"מ בשעה . CTA מוח תקין ?. מה יש לעשות כעת (the question as examined, in Hebrew)
- א. לאשפז ולהתחיל מיד טיפול ב- IV methylprednisolone 1000mg/day
- ב. לשחרר ללא טיפול למעקב ראומטולוגי אמבול טורי
- ג. להתחיל prednisone 40-60 מ"ג ליום ולתאם ביופסיה טמפורלי ת תוך שבועיים
- ד. לאשפז לצורך ביופסיה טמפורלית ללא התחלת טיפול
In the full version for subscribers
- A simple algorithm
- Memory aid
- Exam traps
- 5 review cards
- Enough for IMLE
The full version is open to subscribers of: Public health, Medical licensing exam (IMLE), Family medicine, Internal medicine.
Subscriber? Read the full topic Plans and prices 25 free questions
Sources
- Gordis, Epidemiology, 6th ed., 2019 — Assessing the validity and reliability of diagnostic and screening tests
- Fletcher, Fletcher & Fletcher, Clinical Epidemiology: The Essentials, 6th ed., 2020 — Diagnosis
- Landis & Koch, The measurement of observer agreement for categorical data, 1977 — Biometrics – kappa interpretation
Sources are given for further reading; the text is written afresh and not copied from them.
Reviewed by the Examqo AI Professor · 2026-11-06. The text may contain errors; when in doubt, check the source.