The New York Academy of Medicine Teaching Evidence Assimilation for Collaborative Healthcare New York, August 8, 2013 Yngve Falck-Ytter, MD, AGAF for the GRADE team Associate Professor, Case Western Reserve University, Case & VA Medical Center Chief, Gastroenterology & Hepatology, VA Medical Center, Cleveland 1

How did we make clinical decision?  If the basis was not evidence – what was it?  Expert recommendations 2

Institute of Medicine  March 2011 report: “Clinical Practice Guidelines We Can Trust” 1. Establishing transparency 2. Management of conflict of interest 3. Guideline development group composition 4. Evidence based on systematic reviews 5. Method for rating strength of recommendations 6. Articulation of recommendations 7. External review 8. Updating 3

Quality of CPG based on IOM criteria  169 oncology CPGs evaluated ( )  60% published after 2007  Not a single CPG met all 8 IOM criteria 4 Reams et al. Journal of Clinical Oncology 2013

Quality of CPG based on IOM criteria 5 2. COI 1. Transparency 3. Group composition 4. Based on SR 5. Rating recs 7. Ext. review 8. Updating 6. Wording

6 Before GRADE Level of evidence I II III IV V Source of evidence SR, RCTs Cohort studies Case-control studies Case series Expert opinion A Grades of recomend. B C D

7 Before GRADE Level of evidence Ia Ib II III IV V Source of evidence Meta-analysis RCTs Cohort studies Case-control studies Case series Expert opinion A Grades of recomend. B C D

So what is quality of evidence?  Confidence in evidence  Confidence in the evidence of benefits  Confidence in the evidence of downsides  Confidence in the evidence in the balance  Recognizing that not all outcomes are equal 8

Importance of outcomes 9 Intermediate outcomes Positive hepatitis B core antibody Amnestic response to re-challenge Loss of protective surface antibody Question (PICO) Should health care worker receive booster vaccination vs. not? Final health outcomes Mortality Liver cancer Liver cirrhosis Chronic hepatitis B infection Acute symptom. infection

I B IIVIII A grading system needs to be outcome-centric Quality Old system Outcome #1 Outcome #2 Outcome #3 GRADE 10

Grades of Recommendations Assessment, Development and Evaluation 11

70+ Organizations

Where GRADE fits in Prioritize problems, establish panel Find/appraise or prepare: Systematic review Searches, selection of studies, data collection and analysis (Re-) Assess the relative importance of outcomes Prepare evidence profile: Quality of evidence for each outcome and summary of findings Guidelines: Assess overall quality of evidence Decide direction and strength of recommendation Draft guideline Consult with stakeholders and / or external peer reviewer Disseminate guideline Implement the guideline and evaluate GRADE 13

GRADE expands quality of evidence determinants Methodological limitations Inconsistency of results 14 Risk of bias Allocation concealment Failure of blinding Losses to follow-up Incomplete reporting Indirectness of evidence Imprecision of results Publication bias

15 GRADE: Quality of evidence Although quality of evidence is a continuum, we suggest using 4 categories:  High  Moderate  Low  Very low For guidelines: The extent to which our confidence in an estimate of the treatment effect is adequate to support a particular recommendation.

Determinants of quality  RCTs start high  Observational studies start low 16

Quality of evidence: beyond risk of bias Definition: The extent to which our confidence in an estimate of the treatment effect is adequate to support a particular recommendation Methodological limitations Inconsistency of results Indirectness of evidence Imprecision of results Publication bias Risk of bias: Allocation concealment Blinding Intention-to-treat Follow-up Stopped early Sources of indirectness: Indirect comparisons Patients Interventions Comparators Outcomes 17

18 Quality assessment criteria Lower if… Quality of evidence High Moderate Low Very low Study limitations (design and execution) Inconsistency Indirectness Imprecision Publication bias Higher if… What can raise the quality of evidence? Study design RCTs  Observational studies 

BMJ 2003;327:1459–61 19

20

Question to the audience A. High B. Moderate C. Low D. Very low You review all colonoscopies for average risk colon cancer screening in your health system and document a percentage of patient who developed a perforation after the procedure (evidence of free air on imaging). No comparison group without colonoscopy available. Rate the quality of evidence for the outcome perforation: 21

22 Quality assessment criteria Lower if… Quality of evidence High Moderate Low Very low Study limitations (design and execution) Inconsistency Indirectness Imprecision Publication bias Higher if… Study design RCTs  Observational studies  Large effect (e.g., RR 0.5) Very large effect (e.g., RR 0.2) Evidence of dose-response gradient All plausible confounding… …would reduce a demonstrated effect …would suggest a spurious effect when results show no effect

GRADE evidence profile: HCC associated with HCV eradication Morgan R, Baack B, Smith B, Yartel A, Pitasi M, Falck-Ytter Y. Ann Intern Med. 2013;158: Summary of findingsAnticipated absolute effects Studies (N) IssuesNon- response Viral eradication Relative effects HCCs with no Rx HCCs after viral eradication Quality of evidence Outcome: Hepatocellular carcinoma (importance: critical for decision making) 12 obs. studies (25,906) No serious* risks of bias; large effect 990/16, % 145/9, % RR % CI: 0.18 – o.31 All stages of fibrosis: 17 HCCs per 1, fewer HCCs per 1,000 [-12; -15]   due to large effect * Most studies controlled for baseline liver disease severity (for example, presence of cirrhosis) and other important confounders, such as hepatitis B virus infection. Advanced fibrosis: 33 HCCs per 1, fewer HCCs per 1,000 [-18; -26]

From evidence to recommendations 24 RCT Obser- vational study High level recommen- dation Lower level recommen- dation Old system Quality of evidence Balance between benefits, harms & burdens Patients’ values & preferences GRADE

Values and preferences  Implicit value judgments in recommendations  Trade-offs: example prevention of VTE in surgery  Thrombotic events  Deep vein thrombosis, pulmonary embolism  Bleeding events  Gastrointestinal bleeds, operative site bleeds  Inconvenience of injections  Variability in values and preferences

Case  77 y/o patient with atrial fibrillation, mild CHF, HTN, DM and history of stroke (fully recovered)  Meds: warfarin, antihypertensives, statin, glyburide  Admitted with nausea/vomiting, then hematemesis; INR 2.5; 1 U blood transfused; EGD: no active bleed, possible Mallory Weiss  This is his second major bleed since he started warfarin one year ago

CHADS2 score

Acceptable additional bleeds?  Study: Patients at high risk for atrial fibrillation and high risk of stroke (h/o CHF/MI); internists and cardiologists  Warfarin decreases risk at cost of increased GI bleeds  Without treatment 100 patients will suffer:  12 strokes (six major, six minor), 3 serious GI bleeds in 2 years  Warfarin would decrease strokes in 100 patients to 4 per 2 years (8 fewer strokes, 4 major, minor)  How many additional bleeds would you accept in 100 patients over a year, and still be willing to administer/take warfarin? Slide courtesy of: G. Guyatt; Study: Devereaux et. al., 2001

Slide courtesy: G. Guyatt; Study: Devereaux et. al., 2001

Strength of recommendation “The strength of a recommendation reflects the extent to which we can, across the range of patients for whom the recommendations are intended, be confident that desirable effects of a management strategy outweigh undesirable effects.”  Understanding values & preferences necessary to trade-off benefits and downsides  Values and preferences should ideally be informed by systematic reviews, but evidence is often sparse 30

Example recommendation ACCP AT9 recommendation: In patients undergoing major orthopedic surgery (e.g., total hip replacement), we suggest the use of LMWH in preference to the other agents. Patients who place a high value on avoiding bleeding complications and a low value on its inconvenience are likely to choose a compression device (IPCD) over the drug options. 31

4 determinants of the strength of recommendation Factors that can weaken the strength of a recommendation Explanation  Lower quality evidenceThe higher the quality of evidence, the more likely is a strong recommendation.  Uncertainty about the balance of benefits versus harms and burdens The larger the difference between the desirable and undesirable consequences, the more likely a strong recommendation warranted. The smaller the net benefit and the lower certainty for that benefit, the more likely is a weak recommendation warranted.  Uncertainty or differences in patients’ values The greater the variability in values and preferences, or uncertainty in values and preferences, the more likely weak recommendation warranted.  Uncertainty about whether the net benefits are worth the costs The higher the costs of an intervention – that is, the more resources consumed – the less likely is a strong recommendation warranted. 32

Monthly cost of cancer drugs of 12 cancer drugs approved by the FDA in 2012 cost more than $100,000 / year

Example: ipilimumab (metastatic melanoma) Estimated absolute effects at 2 years Outcome Relative effect (95% CI) control (per 1000) ipilimumab (per 1000) Difference (per 1000) Certainty of the effect Death HR 0.68 (0.55 to 0.85) fewer deaths (49 to 202 fewer)  High Serious immune- related AEs RR 7.37 (4.42 to 12.3) more SAEs (134 to 441 more)  High Quality of life Resource use Ipilimumab: $120,000 (12 weeks induction (4 injections)); Dacarbazine: $400 Increase in median survival: 3.6 mo Incremental cost-effectiveness: ~$400, per life year; $ per QALY???? 34

Developing recommendations 35

Implications of a strong recommendation  Population: Most people in this situation would want the recommended course of action and only a small proportion would not  Health care workers: Most people should receive the recommended course of action  Policy makers: The recommendation can be adapted as a policy in most situations 36

Implications of a conditional recommendation  Population: The majority of people in this situation would want the recommended course of action, but many would not  Health care workers: Be prepared to help people to make a decision that is consistent with their own values/decision aids and shared decision making  Policy makers: There is a need for substantial debate and involvement of stakeholders 37

Systematic review Guideline development PICOPICO Outcome Formulate question Rate importance Critical Important Critical Less important Create evidence profile with GRADEpro Summary of findings & estimate of effect for each outcome Rate overall quality of evidence across outcomes based on lowest quality of critical outcomes RCT start high, obs. data start low 1.Risk of bias 2.Inconsistency 3.Indirectness 4.Imprecision 5.Publication bias Grade down Grade up 1.Large effect 2.Dose response 3.Confounders Rate quality of evidence for each outcome Select outcomes Very low Low Moderate High Formulate recommendations: For or against (direction) Strong or weak (strength) By considering:  Quality of evidence  Balance benefits/harms  Values and preferences Revise if necessary by considering:  Resource use (cost) “We recommend using…” “We suggest using…” “We recommend against using…” “We suggest against using…” Outcomes across studies 38

What GRADE isn’t  Not another “risk of bias” tool  Not a quantitative system (no scoring required)  Not eliminate COI, but able to minimize  Not “expensive”  Builds on well established principles of EBM  Some degree of training is needed for any system  Proportionally adds minimal amount of extra time to a systematic review

40

Summary  Using GRADE enables organizations to produce methodologically rigorous recommendations  It’s sensible, transparent, and systematic and fulfills requirements for use in performance measure production (e.g., NQF, PCPI)  International standardization facilitates direct comparisons across organizations and has the potential to reduce redundancy in efforts 41