Κριτική αξιολόγηση τυχαιοποιημένης κλινικής δοκιμής Άρης Λιάκος, MD MSc Ειδικευόμενος Παθολογίας, Υποψήφιος Διδάκτωρ

Transcription

1 Αριστοτέλειο Πανεπιστήμιο Θεσσαλονίκης Σχολή Επιστημών Υγείας Τμήμα Ιατρικής Πρόγραμμα Μεταπτυχιακών Σπουδών «Ερευνητική Μεθοδολογία στην Ιατρική και τις Επιστήμες Υγείας» Μάθημα: Ιατρική Βασισμένη στην Τεκμηρίωση Κριτική αξιολόγηση τυχαιοποιημένης κλινικής δοκιμής Άρης Λιάκος, MD MSc Ειδικευόμενος Παθολογίας, Υποψήφιος Διδάκτωρ

2 Connolly et al. N Engl J Med 2009;361(12):

3 1. Did the trial address a clearly focused issue? HINT: An issue can be focused in terms of The population studied The intervention given The comparator given The outcomes considered Critical Appraisal Skills Programme (2017) CASP Randomised Controlled Trial Checklist [online]

4 2. Was the assignment of patients to treatments randomised? HINT: Consider How this was carried out? Was the allocation sequence concealed from researchers and patients? Critical Appraisal Skills Programme (2017) CASP Randomised Controlled Trial Checklist [online]

5 Randomisation Minimises selection bias Types Simple Blocked randomization (e.g. AABB, BBAA, ABAB, ABBA, BAAB, e.t.c.) Stratified randomisation

6 3. Were all of the patients who entered the trial properly accounted for at its conclusion? HINT: Consider Was the trial stopped early? Were patients analysed in the groups to which they were randomised? Critical Appraisal Skills Programme (2017) CASP Randomised Controlled Trial Checklist [online]

7 Lost to follow up Losses to follow-up Data missing at random vs. data not missing at random Should be minimal Reasons should be provided Intention-to-treat or per-protocol (completers) analysis Appropriate imputation methods for missing data should be used True ITT Per-protocol or treatment-received Analyse everybody Analyse those who did comply and did finish Did not comply to treatment ecause we want to estimate the effectiveness of the intervention nder real circumstances ΙΤΤ and analysis not the efficacy of the chemical ubstance or the nature of the intervention Per-protocol analysis

8 Trials stopped early for benefit Α Β C Early stop for benefit True effect No effect Interim analyses Final analysis Bassler et al. JAMA 2010;303(12):1180-7

9 4. Were patients, health workers and study personnel blind to treatment? HINT: Think about Patients? Health workers? Study personnel? Critical Appraisal Skills Programme (2017) CASP Randomised Controlled Trial Checklist [online]

10 Allocation concealment vs. blinding Allocation concealment sealed, opaque, sequentially numbered envelopes IVRS web-based computer system independent pharmacist Blinding participants (single) study personnel (double) outcome assesors (triple) study statistician (quadra) minimises ascertainment bias, might be impractical

11 5. Were the groups similar at the start of the trial? HINT: Look at Other factors that might affect the outcome such as age, sex, social class Critical Appraisal Skills Programme (2017) CASP Randomised Controlled Trial Checklist [online]

12 6. Aside from the experimental intervention were the groups treated equally? Critical Appraisal Skills Programme (2017) CASP Randomised Controlled Trial Checklist [online]

13 7. How large was the treatment effect? HINT: Consider What outcomes were measured? Is the primary outcome clearly specified? What results were found for each outcome? Critical Appraisal Skills Programme (2017) CASP Randomised Controlled Trial Checklist [online]

14 8. How precise was the estimate of the treatment effect? HINT: Consider What are the confidence limits? Critical Appraisal Skills Programme (2017) CASP Randomised Controlled Trial Checklist [online]

15 9. Can the results be applied in your context (or to the local population)? HINT: Consider whether Do you think that the patients covered by the trial are similar enough to the patients to whom you will apply this?, if not how do they differ? Critical Appraisal Skills Programme (2017) CASP Randomised Controlled Trial Checklist [online]

16 10. Were all clinically important outcomes considered? HINT: Consider Is there other information you would like to have seen? If not, does this affect the decision? Critical Appraisal Skills Programme (2017) CASP Randomised Controlled Trial Checklist [online]

17 11. Are the benefits worth the harms and costs? HINT: Consider Even if this is not addressed by the trial, what do you think? Critical Appraisal Skills Programme (2017) CASP Randomised Controlled Trial Checklist [online]

18 Subgroup analyses Looking for effect modification Exploratory in nature (hypothesis generating) Credibility based upon: Is the subgroup variable a characteristic measured at baseline or after randomisation? Was the hypothesis specified a priori? Was the subgroup effect one of a small number of hypothesised effects tested? Is the size of the subgroup effect large? Is the interaction consistent across closely related outcomes within the study? Is the significant subgroup effect independent? Sun et al. BMJ 2012;344

19 Sample size calculation If the sample size it too small: a statistically significant difference between the study arms cannot be detected (wide confidence intervals) resources will be wasted it is unethical for participants to enroll in a trial that is not adequately powered to answer the research question If the sample size is too large: resources will be unnecessarily wasted the research proposal is unlikely to receive funding

20 Sample size calculation (continued ) Sample size is usually calculated based on the primary outcome Sample size depends on: Level of significance (type I error rate) Power of the study (type II error rate) The minimally clinically important difference Standard deviation in the population The underlying event (rate) prevalence in the population Other parameters to consider for sample size calculation: The expected drop-out rate The allocation ratio Secondary outcomes or subgroup analyses

21 Non-inferiority trial design Test whether an experimental treatment is not substantially worse than a control treatment Incremental benefits of newly developed treatments may be only marginal or use of placebo unethical Null hypothesis: effects of treatments differ by no more than a tolerable amount (i.e. equivalence margin) Including dropouts in the analysis (ITT analysis) tends to bias the results toward equivalence

22 Tips for interpreting medical trials Do not rely solely on the pathophysiologic mechanism of action Be critical to surrogate endpoints Be critical to comparisons against placebo For active comparators check that therapeutic levels have been used Beware of publication bias Adapted from Greenhalgh Τ. How to read a paper. 2001

23 Tips for interpreting medical trials (continued ) Always look for confidence intervals in the results (statistical versus clinical significance) Respect the hierarchy of the evidence Be critical to claims presented by key opinion leaders Adapted from Greenhalgh Τ. How to read a paper. 2001