隊列研究和CER-AHRQ

1、Copyright H Photography 2006 iSPragmatic cohort studies and comparative effectivenessEric S Johnson, PhDKaiser PermanentesCenter for Effectiveness and Safety ResearchAHRQs Effective Healthcare Program:Scientific Resource Center June 18, 2010 UC Davis School of Medicine1Copyright Matt Knannlein 2005

2、iStockPThe risk of death associated with the use of conventional versus atypical antipsychoticsamong elderly patientsSebastian Schneeweiss, Soko Setoguchi, Alan Brookhart, Colin Dormuth, Philip S WangCanadian Medical Association Journal 2007;176:627-322Copyright Matt Knannlein 2005 iStockPBMJs PICO

3、format: Study questionDo conventional antipsychotic drugs pose risks equal to orgreater than those associated with newer, atypical drugsin an elderly population? 3Copyright Matt Knannlein 2005 iStockPBMJs PICO format: Design, intervention and controlWe conducted a retrospective cohort study of elder

4、ly patients with a diagnosis of dementia who started a conventional or an atypical antipsychotic drug. 4Copyright Matt Knannlein 2005 iStockPBMJs PICO format: Participants and settingWe followed 37,241 patients 65 years or older who started an antipsychotic drug between 1996 and 2004 and wereresiden

5、ts of British Columbia. 5Copyright Matt Knannlein 2005 iStockPBMJs PICO format: Outcomes and their timingWe compared the 180-day all-cause mortality rate for patientswho started conventional versus atypical antipsychotic drugs. 6Copyright Matt Knannlein 2005 iStockPBMJs PICO format: Results and role

6、 of chanceOf the 24,359 (control cohort) patients who started an atypical antipsychotic drug, 9.6% died by 180 days. In adjusted analyses, patients who started a conventional drug were 32% more likely to die (hazard ratio=1.32; 95% CI, 1.23 to 1.42)an excess of 3.5 deaths per 100 patients.7Copyright

7、 Matt Knannlein 2005 iStockPA pragmatic-explanatory continuum indicator summary (PRECIS): a tool to help trial designersKevin Thorpe, Merrick Zwarenstein, Andrew Oxman, and othersJ Clin Epidemiology 2009;62:464-75.“Pragmatic trials seek to answer the question: Does the intervention work under usual

8、conditions. ”8Copyright 2007 Robert Churchill iStockPThe pragmatic to explanatory trials continuum* Flexibility of comparisoninterventionPractitionerExpertise(experimental)Flexibility of theexperimentalinterventionEligibilitycriteriaPrimaryanalysisPractitioneradherenceParticipantcomplianceOutcomesFo

9、llow-upintensityPractitionerexpertise(comparison)E* J Clin Epidemiology 2009;62:464-759Copyright 2007 Robert Churchill iStockPA pragmatic cohort study: new users10Copyright Matt Knannlein 2005 iStockPEvaluating medication effects outside of clinical trials:New-user designsWayne RayAm J Epidemiol 200

10、3;158:915-20“Prevalent users can introduce two types of biases:Underascertainment of events that occur early in therapy Inability to control for predictors that may have been altered by the study drugs.”Prevalent users also preclude meaningful propensity scores.11Copyright 2007 Robert Churchill iSto

11、ckPValid comparisons with usual care?12Copyright 2006 Ethan Myerson iStockPHow might we design a score to predict the usefulness of pragmatic cohort studies?How well does the cohort answer decision-makers questions?Inferiority or non-inferiority?How closely does the cohort resemble a pragmatic trial

12、? Design? Analysis?How easy is the cohort for systematicreviewers to GRADE and synthesize?An unintended outcome, a harm?. 13Copyright 2006 Ethan Myerson iStockP“A national CER programshould promote the professional ethos that places the interests ofpatients and the largercommunity above all othercon

13、siderations.”Harold Sox and Sheldon GreenfieldAnnals of Intern Med 2009;151:203-5Comparative effectivenessresearch: a report from theInstitute of Medicine 14Copyright Juan Monino 2006 iStockP“He wants to look at his plate,see a steak, and say,I like steak! ”“Give people what they want,then later you

14、 can give themwhat you want.” Pascal in Big Night (1996)15Copyright Matt Knannlein 2005 iStockPVariation in the risk of suicide attempts andcompleted suicides by antidepressant agentin adultsSebastian Schneeweiss, Amanda Patrick, Daniel Solomon, Jyotsna Mehta,Colin Dormuth, Matthew Miller, Jennifer

15、Lee, Philip S WangArchives of General Psychiatry 2010;67:497-50616Copyright Matt Knannlein 2005 iStockPDesign, intervention and control:Advanced strategies for propensity scores“After plotting and comparing the distribution of propensity scoreswe truncated our study population to the area of overlap

16、. Subjects with non-overlapping propensity scores mayhave absolute clinical indications or contraindications for aparticular treatment.”17Copyright Matt Knannlein 2005 iStockPDesign, intervention and control:Advanced strategies for propensity scores“We constructed a high-dimensional propensity score based on the top 500 identified variables in addition to the variablesidentified by usWe consider results from this analysis to be the most completely adjusted ”18