210 likes | 261 Views
Diagnostic Cases. Goals & Objectives . Highlight Bayesian and Boolean processes used in classic diagnosis Demonstrate use/misuse of tests for screening vs. diagnosis Have fun while learning about some common clinical questions. Seven standards for Tests. Spectrum composition
E N D
Goals & Objectives • Highlight Bayesian and Boolean processes used in classic diagnosis • Demonstrate use/misuse of tests for screening vs. diagnosis • Have fun while learning about some common clinical questions
Seven standards for Tests • Spectrum composition age distribution, sex distribution, presenting clinical symptoms and/or disease stage, and eligibility criteria for study subjects. • Pertinent subgroups • Avoidance of workup bias • Avoidance of review bias • Precision of results for test accuracy • Presentation of indeterminate test results • Test reproducibility
From Bandiolier http://www.jr2.ox.ac.uk/bandolier/band26/b26-2.html Out of total= 7 standards recommended Year of article publication
Case #1 Strep Throat
The 9 year old, if he had NO rash, would get most benefit from testing. The teacher is benefited mostly by a positive test. How do you tell a “carrier” state from a disease causing strep?
A Bayes Rule of Thumb: Tests work best when Pretest Probability is 50:50
Screening Principles • Is the problem serious, and do patients care about it? • Is the screening test accurate? • Is the “gold standard” comparison reliable? • Is the positive predictive value acceptable? • Does early detection of the disease improve outcomes? • Is screening or treatment benign (i.e. not harmful)? • Does screening do more good than harm? • In a world of limited resources, is screening cost effective? • Absolutely effective compared to natural hx of disease? • Relatively effective compared to using resources to find/treat other problems?
Chief Complaint Sadie Blue is a 22 year old female. Her chief complaint is “no energy". History of Present Illness She reported: enjoys interaction with opposite sex none of the time | depressed most of the time | feel best in morning some of the time | normal thinking none of the time | full life some of the time | irritable most of the time | decisive none of the time | restless a good part of the time | hopeful none of the time | useful none of the time | crying spells a good part of the time | enjoying activities none of the time. She denied: suicidal ideation some of the time. Past, Family, and Social History Social History Activities for Daily Living History of: normal activities none of the time. Review of Systems Constitutional She reported: eating as much as before some of the time | weight loss a good part of the time | fatigue most of the time. Cardiovascular She denied: palpitations some of the time. Gastrointestinal She reported: constipation a good part of the time. Neurological She reported: dyssomnia most of the time. Self-assessment Scales Title: Zung Depression Scale Description: This 14-item scale for depression is a classic in self-rating scales. William Zung at Duke University published this early scale for patient use in 1965. Valued for its brevity, it remains a useful screening tool for depression. Patient Score: 65 - Moderate to Marked Scoring Key and Interpretation: 0 - 50 : Normal 51 - 60 : Minimal to Mild 61 - 69 : Moderate to Marked 70 - 999 : Severe to Extreme Reference: Zung, W.W.K.: A self-rating depression scale. Archives of General Psychiatry, 1965; 12:63-70. Depression Case
What does this mean? Title: Zung Depression Scale Description: This 14-item scale for depression is a classic in self-rating scales. William Zung at Duke University published this early scale for patient use in 1965. Valued for its brevity, it remains a useful screening tool for depression. Patient Score: 65 - Moderate to Marked Scoring Key and Interpretation: 0 - 50 : Normal 51 - 60 : Minimal to Mild 61 - 69 : Moderate to Marked 70 - 999 : Severe to Extreme Reference: Zung, W.W.K.: A self-rating depression scale. Archives of General Psychiatry, 1965; 12:63-70.
Mammography & CAD My wife recently had a mammogram. She came home and said, "They asked me if I wanted to pay $25 more to have a computer help read my mammogram. I told them 'No, that's the doctors job!'. Was that the right thing to do?"
Mammography & CAD Radiologist Alone CAD Alone Combined R+CAD
Figure 1. ROC curves and sensitivity and specificity data obtained from the interpretation of 104 mammograms by 10 radiologists. A cluster of microcalcifications was present in all cases; 46 cancers and 58 benign lesions were confirmed at biopsy. The effect of a computer aid was tested; it provided an estimate of the likelihood that microcalcifications were due to a malignancy. Sensitivity and specificity results were based on the radiologists’ recommendations for biopsy or follow-up. The ROC curves were based on the radiologists’ diagnostic confidence.
Summary • For uncommon illnesses (screening, like breast cancer) there will be lots of false positives. • Apply the test correctly, to the correct population • “Clinical judgment” means you figure out which population the patient belongs to, before applying the test (i.e. good pretest probability) • Good tools for pretest probability are hard to find: use the ones we have well! • Watch out for back end costs- complications and death from testing, anaphylaxis from antibiotics, social stigma from psych diagnoses, etc.
Reference • How to Read a Paper: Papers that Report Diagnostic or Screening Tests. BMJ 1997: 315: 540-543 (August 30). • Available on Internet, full text.