Overview. Why use tests? INTRODUCTION TO TEST EVALUATION RESEARCH

Transcription

1 Overview INTRODUCTION TO TEST EVALUATION RESEARCH. Tests 2. Evaluation of Tests RCT 3. Target Condition 4. Test Accuracy Studies 5. Systematic Reviews of Test Accuracy 6. Coda Patrick MM Bossuyt. Tests Why use tests? Diagnosis Monitoring course disease Selecting therapy Following effects of therapy Determining drug levels or drug effects Evaluate Health or Fitness Screening Case Finding

2 2. Evaluate Tests Medical Test Information Can I trust the information from this test? Medical Test Information Medical Test Information Can I trust that this information improves patients health? Can I trust that this information tells me something about the patients health? Patient 2

3 Randomized Clinical Trial Medical Test RCT Active TEST Population Study Group Randomize Population Study Group Randomize Control Control Medical Test RCT Medical Test RCT TEST TEST Population Study Group Randomize Population Study Group Randomize No TEST TEST B Meta Analysis Screening RCT Combination antimicrobial susceptibility testing for acute exacerbations in chronic infection of Pseudomonas aeruginosa in cystic fibrosis Valerie Waters, Felix Ratjen Year: 28 Screening for colorectal lcancer using the faecal occult blood test, t Hemoccult P Hewitson, P Glasziou, L Irwig, B Towler, E Watson Year: 27 Urodynamic investigations for management of urinary incontinence in children and adults CMA Glazener, MC Lapitan Year: 22 Relative Risk CRC Mortality:.84 (95% CI:.77 to.93) BMJ 998;37:

4 Lancet 996;348:473 UK RCT: CRC incidence & mortality Rate (/ pyrs) RR Screening Control CRC CRC mortality Total mortality (.98.3) Median follow-up 7 8 years Lancet 996;348:475 Urodynamic investigations for management of urinary incontinence in children and adults Objectives The objective of this review was to discover if treatment according to a urodynamic based diagnosis led to clinical improvements in urinary incontinence outcomes, compared to treatment based on history and examination. Three small trials were found, which included 84 people, although information was only available for 28 participants. There was not enough evidence to determine whether these tests lead to better outcomes. There was some evidence that urodynamic testing increased the number of people prescribed drug treatments or treated by surgery, but it was not known whether this resulted in less incontinence or a better quality of life. More research is needed, in which people are randomised to having treatment decisions based on either their symptoms and examination alone, or the extra information provided by urodynamic tests. Lancet 996;348:468 4

5 Tubal integrity testing 5

6 Hum Reprod. 26 2:227-3 Hum Reprod. 26 2:227-3 RCT Medical Test RCT Medical Test Test Positive Left Test A Result Study Group Randomize Study Group Randomize Negative Positive Right Left Test B Result No Test Negative Right Bossuyt et al; Lancet 2 6

7 RCT of Testing Best evidence of effectiveness Rare Usually need large sample sizes Need protocol Need patient outcomes that matter Medical Test Test Results 3. Target Condition Clinical Decision Treat Wait Patient Medical Test Test Results Medical Test Test Results Cognition Emotion Clinical Decision Clinical Decision Behavior Treat Wait Treat Wait Patient Patient 7

8 Spectrum of disease Venous Thromboembolism Probability of Positive Test Result Benefits of Treatment Spectrum of Disease ` AJR:74, June 2 Roles of tests 4. Test Accuracy Bossuyt et al; BMJ 26 8

9 Bladder Tumor Markers (BTM) Diagnostic Accuracy To optimize monitoring of tumor recurrence and progression, without incurring more invasive and expensive medical tests Index Test Reference Standard Diagnostic Accuracy Study The Results Series of Patients Reference Standard Index Test Target Condition Other Condition Reference standard Cross classification Positive Negative Bladder Cancer close follow up due to the significant risk of tumor recurrence. Cytology: central test Efficient Non invasive Inexpensive But imperfect Cystoscopy 9

10 Bladder Tumor Markers (BTM) Example To optimize monitoring of tumor recurrence and progression, without incurring more invasive and expensive medical tests Diagnostic Accuracy Study The Results 5 patients Cystoscopy BTA stat t BCR No BCR Cystoscopy Cross classification Positive 7 96 Negative Measures of Diagnostic Test Accuracy Sensitivity & Specificity 54 % 74 % PPV & NPV 43 % 8 % Likelihood Ratios Diagnostic Odds Ratio 3.2 BCR no BCR Positive 7 96 Negative

11 Message Measures of Diagnostic Test Accuracy 5. Systematic Reviews are test characteristics, fixed test properties Why systematic reviews Extensive/exhaustive search Critical appraisal Meta analysis Increased precision Explore variability. Focused question Systematic Review: Question Elements What is the diagnostic accuracy of Systematic Review: Question Elements What is the diagnostic accuracy of <Index test> [versus <comparator>] for detecting <target condition> in <patient description> BTA stat test versus cytology for detecting recurrent disease in patients with bladder cancer

12 Sensitivity ROC space Identification & selection studies Specificity 3. Quality Assessment Differential Verification Bias 4. Meta analysis 2

13 3 Study results Paired Forest Plots ROC space y Specificity Sensitivity ROC space y Specificity Sensitivity ROC space y Specificity Sensitivity ROC space y Specificity Sensitivity

14 Summary Point in ROC space Summary Point in ROC space Summary Point in ROC space Summary Point in ROC space Summary Curve in ROC space Paired studies: cytology & BTA Squares: Diamonds: cytology BTA 4

15 5. Interpretation Museum Boijmans van Beuningen, Rottrrdam 5

16 6