Transcript Presentation 6: systematic review talk CER

Shari D. Bolen, MD, MPH
MetroHealth/CWRU
Assistant Professor of Medicine,
Epidemiology & Biostatistics
I have no COI





Define a systematic review and metaanalysis
Explain the steps and timeline for a high
quality systematic review
Discuss uses of systematic reviews for CER
Discuss novel areas for growth in
systematic reviews conducting CER
Potential funding sources


Systematic review: a summary of existing
research using reproducible methods
Meta-analysis: a statistical approach to
combining data from individual studies
Define question
Establish eligibility criteria
Search literature & review citations
Extract data
Evaluate study quality & applicability
Summarize & synthesize

Determine priorities
◦ Identify areas of controversy (do guidelines or
experts conflict?)

Formulate in specific measurable terms
◦ Use a structured approach to frame each question
(PICOTS)

Be specific about:
◦ Population: Adults with Type 2 diabetes
◦ Intervention: TZDs, biguanides, SUs, alphaglucosidase inhibitors; meglitinides
◦ Comparison: head to head and indirect;
monotherapy
◦ Outcomes: intermediate, long term, and safety
◦ Time: studies>3 months
◦ Setting: no limits

Select highest yield sources

Consider other sources (CINAHL, ERIC,
CancerLit, FDA, etc…)
◦ MEDLINE®/PubMed
◦ EMBASE®
◦ Cochrane database of reviews & trials
◦ Use sampling to estimate incremental yield

Use hand searches for quality control
◦ Citations in eligible articles
◦ Table of contents of relevant journals
◦ Query experts



Develop strategy in Pubmed then translate to
other websites
Use literature on search strategies (BMJ)
For diabetes medications, we used the terms:
type 2 diabetes and the individual generic
and brand names of the medications
Title review (n=7,563)
Title review: 5,157 excluded
Abstract review (n=2,406)
Abstract review: 1,972 excluded
Full-text review (n=434)
Full-text review: 218 excluded
216 included articles








Extract data using standardized data extraction
forms
Sequential or Independent dual review
Discuss and resolve discrepancies
Use established instruments for quality
assessment (Jadad, Strobe)
Assemble evidence tables
Summarize results
Perform meta-analyses if indicated
GRADE strength of the evidence


Often takes 1-3 years to conduct a high
quality systematic review
Budget: can do unfunded if fairly small; large
reviews often cost $250,000 per year
A1c Results: Systematic Review CER for Guidelines
Bennett W L et al. Ann Intern Med 2011;154:602-613
©2011 by American College of Physicians
Systematic Review CER as
“Whistle Blowers”
Expression before talking with GSK lawyers Expression after talking with GSK lawyers
Systematic Review CER as “Whistle Blowers”
Comparators
Pooled OR of
CV risk
(95% CI)
Metformin vs placebo or other med 0.85 (0.69 to
1.05)
SU vs placebo or other med
0.89 (0.71 to
1.11)
Pioglitazone vs placebo or other
med
0.88 (0.78 to
1.00)
Rosiglitazone vs placebo or other
med
1.68 (0.92 to
3.06)



Need long term observational studies to compare
different medications effects on long term and safety
outcomes (after adjusting for selection bias)
Need standard reporting of outcomes in order to
combine studies in meta-analyses
Need to include patients with complications and
comorbid illnesses which are often left out of RCTs



Synthesizing indirect comparisons
Incorporating patient input
Assessing study applicability
Psaty et al. JAMA




AHRQ EPCs
CDC
Organizations such as SGIM, AHA, etc…
NIH possibly (although more challenging)

Systematic reviews are great opportunities
for:
◦ Answering existing CER questions
◦ Setting up future CER studies
◦ Developing novel methodologies for CER,
including incorporation of patient input