Lecture jr

314 views

Published on

0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
314
On SlideShare
0
From Embeds
0
Number of Embeds
2
Actions
Shares
0
Downloads
3
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

Lecture jr

  1. 1. From data to publication Jonas Ranstam PhD
  2. 2. theoryoutcome hypothesis data
  3. 3. theory reasoning study designoutcome hypothesisstatistical analysis data collection data
  4. 4. reasoning study designstatistical analysis data collection
  5. 5. To discuss in this presentationstudy design - observation (case report, survey, epidemiological study) - experiment (phantom, in vitro, in vivo, clinical trial)data collection - registration, monitoring, validation, documentationstatistical analysis - data description, effect estimation, evaluation of bias and uncertaintyreasoning - interpretation of outcome with respect to the limitations imposed by study design, data collection and statistical analysis
  6. 6. Design features (simplified)------------------------------------------------------------------------------------------ Design --------------------------------------------------------Characteristics Experimental Observational------------------------------------------------------------------------------------------Studied effects beneficial harmfulSample size small largeFollow up short longInternal validity better worseMain outcome efficacy effectivenessExternal validity worse better------------------------------------------------------------------------------------------
  7. 7. Design features (simplified)---------------------------------------------------------------------------------------------------------------------- Design ----------------------------------------------------------------------------------Characteristics Experimental Observational----------------------------------------------------------------------------------------------------------------------Data collection from CRF to database from ? to registerStatistical analysis precision oriented validity orientedReasoning multiplicity, missing data, confounding, selection and compliance, superiority, information bias, measurement non-inferiority errors----------------------------------------------------------------------------------------------------------------------
  8. 8. Data collection
  9. 9. DIRECTIVE 2001/20/EC OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL of 4 April 2001 on the approximation of the laws, regulations and administrative provisions of the Member States relating to the implementation of good clinical practice in the conduct of clinical trials on medicinal products for human use4. All clinical trials, including bioavailability and bioequivalence studies, shall be designed, conducted and reported in accordance with the principles of good clinical practice.
  10. 10. ICH-GCP1.24 Good Clinical Practice (GCP)A standard for the design, conduct, performance,monitoring, auditing, recording, analyses, and reportingof clinical trials that provides assurance that the dataand reported results are credible and accurate, and thatthe rights integrity and confidentiality of trial subjects areprotected.
  11. 11. ICH-GCP4.9 Records and ReportsThe investigator should ensure the accuracy,completeness, legibility, and timeliness of all the datareported to the sponsor in the CRFs and I all requiredreports.
  12. 12. ICH-GCP5.5 Trial Management, Data Handling, and RecordKeepingIf data are transformed during processing, it shouldalways be possible to compare the original data andobservations with the processed data.( Audit trail: Documentation that allows reconstruction of the course of events)
  13. 13. ICH-GCP1.6 AuditA systematic and independent examination of trialrelated activities and documents to determine whetherth evaluated trial related activities were conducted, andthe data were recorded, analyzed and accuratelyreported according to the protocol, sponsors standardoperating procedures (SOPs), Good Clinical Practice(GCP), and the applicable regulatory requirement(s).
  14. 14. Obligation to Register Clinical TrialsThe ICMJE defines a clinical trial as any researchproject that prospectively assigns human subjects tointervention or concurrent comparison or control groupsto study the cause-and-effect relationship between amedical intervention and a health outcome. Medicalinterventions include drugs, surgical procedures,devices, behavioral treatments, process-of-carechanges, and the like.
  15. 15. Requirements for observational studiesICMJE - Selection and Description of ParticipantsDescribe your selection of the observational or experimentalparticipants (patients or laboratory animals, includingcontrols) clearly, including eligibility and exclusion criteria anda description of the source population.
  16. 16. Requirements for observational studiesThe STROBE statementDescribe the setting, locations, and relevant dates, includingperiods of recruitment, exposure, follow-up, and datacollection.For each variable of interest, give sources of data anddetails of methods of assessment (measurement).
  17. 17. Statistical analysis and reasoning
  18. 18. Misunderstandings about statistical calculations- They reveal otherwise unknown information about the studied population (sample)- Their purpose is to find statistically significant differences or effects- It is only interesting whether a difference has p<0.05 or “ns”- If a difference is not significant, it does not exist- If a difference is significant, it is practically important- It is important to test all differences, especially for evaluating the success of randomization (in clinical trials) and matching (in observational studies)- Findings can only be published if they are statistically significant
  19. 19. Sampling and measurement variabilityExperiment A What do we know about sampling and measurement variability when an experiment is performed only once? Outcome
  20. 20. Sampling and measurement variabilityExperiment A Experiment A Experiment A Experiment A Experiment A First time Second time Third time Fourth time Fifth time Outcome UncertaintyHad we replicated the experiment several times, the variation hadbeen evident.
  21. 21. Sampling and measurement variabilityExperiment A With only one performance of the experiment, the uncertainty caused by sampling and measurement variation can be evaluated using statistical methodology. Outcome Uncertainty
  22. 22. Sampling and measurement variability Hospital A Hospital B Hospital C Hospital D Hospital E Outcome UncertaintySampling and measurement variability must be taken into account when comparingdifferent entities, otherwise the results cannot be meaningfully interpreted. Politiciansand reporters do generally not understand this.
  23. 23. Observation vs. inferenceFor one particular observed sampleCentral tendency: Mean, Median (statistic)Dispersion: SD, RangeFor the unobserved population of samplesCentral tendency: Mean, Median (parameter)Uncertainty: SEM, confidence interval
  24. 24. Precision and validity of estimates Lower validity Higher validity Higher precision Lower precision
  25. 25. Precision and validityPrecision is often presented using a 95% confidence interval.Like the p-value this is an estimate of the uncertainty related tosampling and measurement variability.What about validity?- Experiments are designed for validity (randomization, blinding, etc.)- Observational studies are analyzed to reduce bias (validity errors).
  26. 26. Confounding bias – crude estimate Birth weightCrudeSmoking effect Non-smokers Smokers
  27. 27. Confounding bias – crude estimate Birth weight Non-smokers SmokersCrudeSmoking effect Gestational age
  28. 28. Confounding bias – adjusted estimate Birth weight Non-smokers SmokersAdjustedSmoking effectCrudeSmoking effect Gestational age
  29. 29. Methodological development in different research areasLevel of proficiency Clinical trials Observational studies Laboratory experiments Time Some time not Today In the not too too long ago far future
  30. 30. Recent developments inthe analysis of observational studies
  31. 31. - Longitudinal analysis using random effects models- Multilevel analysis- Causality models- The development of publication guidelines- Debate on registration of epidemiological studies
  32. 32. Recent developments in clinical trials
  33. 33. - Random effects models for analysis of FAS- Closed test procedure strategies for handling multiplicity issues- Development of superiority, equivalence, and non-inferiority trial designs- The development of ICH guidelines for design and analysis- The development of publication guidelines- Registration of trials in a public register
  34. 34. P-value and confidence interval Information in p-values Information in confidence intervals [2 possibilities] [2 possibilities] p < 0.05 Statistically significant effect n.s. InconclusiveEffect 0
  35. 35. P-value and confidence interval Information in p-values Information in confidence intervals [2 possibilities] [2 possibilities] p < 0.05 Statistically significant effect n.s. InconclusiveEffect 0 Clinically significant effects
  36. 36. P-value and confidence interval Information in p-values Information in confidence intervals [2 possibilities] [6 possibilities] p < 0.05 Statistically but not clinically significant effect Statistically and clinically significant effect p < 0.05 p < 0.05 Statistically, but not necessarily clinically, significant effect n.s. Inconclusive n.s. Neither statistically nor clinically significant effect p < 0.05 Statistically significant reversed effectEffect 0 Clinically significant effects
  37. 37. Superiority, non-inferiority and equivalence Superiority shown Superiority shown less stronglyNon-inferiority not shown Superiority not shown Non-inferiority shown Superiority not shown Equivalence shown Superiority not shown Control better New agent better 0 Margin of non-inferiority or equivalence
  38. 38. Statistical analysis of clinical trialsICH E9 Statistical Principles for Clinical TrialsCPMP Points to consider on multiplicity issues in clinical trialsCPMP Points to consider on adjustment for baseline covariatesCPMP Points to consider on missing dataCPMP Point to consider on switching between superiority and non-inferiority
  39. 39. Recent developments inthe analysis of laboratory experiments
  40. 40. - Bioinformatics (FDR, etc.)- The (slow) development of publication guidelines
  41. 41. Manuscript preparation guidelinesSee http://www.equator-network.org/Clinical trials- CONSORT statement (several)Laboratory experiments (primarily in vivo)- ARRIVE statementObservational studies- STROBE statement

×