Drug information and Guidelines Flashcards
AGREE instrument and Bandolier
tools for evaluating guidelines
Agency for Healthcare Research and Quality (AHRQ)
catalog of guidelines, tools to give you ability to compare guidelines by structured elements (see handout), they NO longer write guidelines
How do you find guidelines?
#PubMed #AHRQ guidelines.gov #Cochrane collaboration #National Institute for Clinical Excellence (British)
Sources of Oncology based Guidelines
#NCCN (both EBM and consensus) #ASCO (EBM) #ONS (specific to RN, both EBM and CBM) #MASCC (both) (Multinational Assoc of Supportive Care in Cancer #ASBMT (EBM) #Int'l Assoc for Hospice and Palliative Care Medicine #ESMO (European) #IDSA #Pediatric Oncology Group of Ontario
Meta-analyses
#statistics on aggregate information #systematic review #way to combine results of multiple trials and to reanalyze #use rigorous methods to prevent bias and erroneous results "garbage in, garbage out" even MORE important here
why would you do a meta-analysis?
#clear clinical question (very impt!!) #address sample size and beta error issues (particularly common in CA studies is too small sample size) #varying, inconsistent results in current trials #address subgroup issues (problematic because sufficient study power DOES NOT apply to subgroups)
what studies do you want to include in a meta-analysis?
SIMILAR #study design #allocation #patient selection #therapies #consistent data points #outcomes
How do you identify studies for meta-analysis?
Find everything!! #search at least 2 databases (ie. PubMed gold std, Web of Science, Embase, Cochrane, Medline, etc.) #search ascendancy (references that your identified article listed) and descendancy (references that cited your identified article POST publication) #unpublished sources (manufacture, FDA, abstracts, US trial registry) #language bias (often limited to English language)
meta-analysis Search terms
#MeSH searching and text word searching #limits (common RCTs only) #date range of search
How are studies selected for inclusion in Meta-analysis?
#NEED TO WORRY ABOUT both study selection bias and patient selection bias #quality assessment
Study assessment for meta-analysis
#multiple reviewers assess all trials #disagreement resolved by consensus #consistent methodology to review and abstract all trials #investigators contacted for missing data #some meta-analysis studies use criteria to assess quality of studies (many formulas available, often assess allocation)
does it make sense to combine studies for meta-analysis?
#patients with similar dz states, interventions, outcomes, study designs #tools to assess: forest plots and heterogeneity
how do you assess for bias in meta-analysis studies?
#selection of pts, allocation #interventions: similar tx #interventions: competing tx #confounding #measurement and observer error #outcomes used #compliance, attrition #duration of tx
forest plots (qualitative) to assess heterogeneity of meta-analysis
Are point estimates highly variable?
Do confidence intervals overlap?
BOX: point estimate (ie. relative risk)
size of box= relative sample size, line whiskers =CI (larger the line, smaller the sample size)->looking to see that boxes are pretty close to show qualitative heterogeneity, looking for outliers. if wild heterogeneity look to studies to dig in and understand
Cochran Q statistic (quantitative) to assess heterogeneity of meta-analysis
based on chi squared statistic, p value less than 0.1 or 0.05 = significant heterogeneity
hypothesis = NOT heterogeneity
Cochran Q statistic (quantitative) to assess heterogeneity of meta-analysis - LIMITATIONS (2pts)
underpowered for studies with few pts, #overpowered for large sample sizes.
I^2 statistic (quantitative) to assess heterogeneity of meta-analysis
variability across studies due to treatment effect #0% variability due to chance alone #0-25% low hetero #25-50% moderate hetero #>50% high #PUSH to included CI's for this stat. (ie. 0-62% would not be ideal but may be reported now as 0%)
fixed effects model - meta-analysis
#assumes one true value #considers sample variation with studies (each pt has SAME weighted value) #use if LOW hetero
random effects model - meta-analysis
#assumes a range of effects #cosiders variation with study AND between studies (pt's could have DIFFERENT weighted values) #use if MOD/HIGH hetero
what is common method to decide which effects model to use? - meta-analysis
#Primarily use fixed effects UNLESS significant hetero THEN Secondarily use random effects #can use different effects model for different outcomes within the SAME study
“leave one out procedure” used in random effects - meta-analysis
assess effect of individual studies by leaving each study out and assessing the results
subgroups analysis- meta-analysis
#step back - does this make sense? #particularly subject to CONFOUNDING BIAS #fishing expeditions (p value penalty) #subgroups may NOT reflect randomization #adjust for multiple comparisons #alpha, beta, power
sensitivity analysis- meta-analysis
#evaluate the impact of different decisions made in the conduct of the study (ie. if both fixed and random effects examined if they give similar answer than sensitivity is robust) #different study designs such as RCT and epidimiology = similar outcomes
funnel plot. What does this assess?
#plot effect size vs sample size #used to assess publication and selection bias #if studies evenly distributed under the funnel - NO publication bias #if studies on ONE side - publication bias people are not excited about publishing neutral studies
example limitations - meta-analysis
using only published data, no quality control on original records, heterogeneity in trial design, differences in chemo regimens, relatively small # of trials and pts
data mutations - meta-analysis
#strategy to be able to calculate NNT off of nominal data #may mean LOSS of data #change continuous data to nominal date (ie. pts who had at least 50% reduction in pain, pts whose bp was in nml range)
Tools for evaluating meta-analysis
provides checklists for what to look for #CASP: critical appraisal skills program #Bandolier #PRISMA Statement
SUPPLEMENTAL MATERIALS TO REVIEW NOT ON THE VIDEO
PubMed searching (subheadings, etc.), other literature searching sites (EMBASE, Cochrane), internet website notes (NCI toxicity criteria, FDA drug and biologic sections (2 sections)), resources for evaluating clinical guidelines, National Guidelines Clearinghouse (different parameters), HIPAA
HIPAA alphabet- PHI
protected health information
HIPAA alphabet- covered entity
or
HIPAA alphabet- BAA
business associates agreement
HIPAA alphabet- TPHA
treatment, payment, and healthcare operations - exclusion principles
HIPAA alphabet- HITECH
amendment that has increased fines, risk, and exposure to organizations
Research and HIPAA
#quality improvement vs research #IRB and HIPPA #PHI disclosures without authorization (reporting an adverse event to the FDA, legally required reporting) #violations and penalties (REVIEW THIS)