David M Thompson Dept of Biostatistics and Epidemiology College of Public Health OUHSC Learning to Practice and Teach EvidenceBased Health Care Fifth Annual Workshop September 2425 2010 ID: 1045126
Download Presentation The PPT/PDF document "“EBHC Statistical Toolkit”" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
1. “EBHC Statistical Toolkit”David M. ThompsonDept. of Biostatistics and EpidemiologyCollege of Public Health, OUHSCLearning to Practice and Teach Evidence-Based Health CareFifth Annual WorkshopSeptember 24-25, 201015th Annual EBHC Workshop 9-24-2010
2. Statistical tools answer questionsby testing hypotheses and generating p-valuesby estimating parameters and generating confidence intervals on those estimates5th Annual EBHC Workshop 9-24-20102
3. Glossaries and online calculators5th Annual Workshop - Learning to Practice and Teach EBHCOUHSC Bird Library - Evidence Based HealthcareDuke - UNC Chapel Hill Intro to EBPEBM calculators at Can. Inst. of Health Research5th Annual EBHC Workshop 9-24-20103
4. Clinical QuestionsEpidemiologyImpact of symptoms and disease on patient or othersEtiologyScreeningDiagnosisTreatment/ManagementPrognosis45th Annual EBHC Workshop 9-24-2010
5. Evaluating (or choosing) statistical tools hinges on the question of interestP PopulationI Intervention, prognostic factor, or exposureC Comparison groupO Primary outcome(Study design)55th Annual EBHC Workshop 9-24-2010
6. Outcome measuresCategoricalBinary disease vs. no diseaseMultilevel and unorderedMultilevel and ordered Disease stage I,II,II,IVOpinion: disagree, neutral, agree5th Annual EBHC Workshop 9-24-20106
7. Outcome measuresNumericDiscrete Counts of events of disease or adverse eventsNumber of apoptotic cellsContinuousHbA1cNatural log of C reactive proteinTime to eventProgression free survivalOverall survival5th Annual EBHC Workshop 9-24-20107
8. OutcomesEBHC glossaries focus on “treatment effects” in studies of an Intervention, Exposure, or Prognostic factorthat presume the outcome is a countable “event”.(http://ktclearinghouse.ca/cebm/glossary/)5th Annual EBHC Workshop 9-24-20108FormulaRisk reductionRisk increaseBenefit increaseRelative|EER - CER|/CERRel. risk reductionRel. risk increaseRel. benefit increaseAbsolute|EER - CER|Harmful or beneficial events per person“Number neededto …”1/ |EER - CER|Persons per harmful or beneficial eventNNTNNHNNT
9. Outcomes measured in other ways require other statistical tools5th Annual EBHC Workshop 9-24-20109
10. Boilerplate“Continuous variables were analyzed using t-tests or, when appropriate, their nonparametric analogs. Associations between categorical variables were assessed using Chi-square tests or, when expected values were small, Fisher’s exact tests.”5th Annual EBHC Workshop 9-24-201010
11. Statistical tools fit the features of the questionP PopulationI Intervention, prognostic factor, or exposureC Comparison groupO Primary outcome(Study design)5th Annual EBHC Workshop 9-24-201011
12. Statistical tools fit the features of the question5th Annual EBHC Workshop 9-24-201012OutcomeComparison group defined by Intervention or ExposurePopulation Covariates Age, Sex Disease Severity Comorbid conditions
13. Features of statistical modelStatistical interaction or “effect modification”Correlated observations of the outcomeMultiple comparisons5th Annual EBHC Workshop 9-24-201013
14. Interaction between marital status and C1 enrollment regarding incidence of infant death5th Annual EBHC Workshop 9-24-201014
15. Certain study designs obtain(and take advantage of) nonindependent (or correlated ) observations of the outcome.Observations can be correlatedtemporallyspatiallyhierarchically5th Annual EBHC Workshop 9-24-201015
16. Statistical tools that appropriatelyhandle correlated observationsRepeated measures analysis of varianceLinear mixed modelsfor numeric outcomesGeneralized linear modelsfor outcomes that are binary, categorical, ordinal, or countsconditional and marginal models5th Annual EBHC Workshop 9-24-201016
17. Multiple comparisonsThe probability of detecting and reporting differences that don’t truly exist accumulates in a study that examines several hypothesis tests.5th Annual EBHC Workshop 9-24-201017
18. 5th Annual EBHC Workshop 9-24-201018
19. The right statistical tool for the question.“Between-group differences in HbA1c were assessed using a mixed regression model that accounted for the study’s repeated and, therefore, correlated measurements on each subject. …”5th Annual EBHC Workshop 9-24-201019
20. “… Hypothesis testing focused on the model’s estimate of group*time interaction to assess whether change in HbA1c over time differed between the treatment groups. …”5th Annual EBHC Workshop 9-24-201020
21. “…The model also produced stratum-specific estimates of the change in HbA1c levels over time (in mg/dL/year) along with 95% confidence intervals.”5th Annual EBHC Workshop 9-24-201021