uc/hh epc comparative effectiveness reviews and evidence- based practice c. michael white, pharmd,...
TRANSCRIPT
UC/HH EPCUC/HH EPC
Comparative Effectiveness Reviews and Evidence-based Practice
C. Michael White, PharmD, FCCP, FCPProfessor of Pharmacy & Interim Dept Head,
Pharmacy PracticeDirector,
University of Connecticut/Hartford Hospital Evidence-Based Practice Center
University of Connecticut School of Pharmacy
UC/HH EPCUC/HH EPC
Faculty Disclosure
C. Michael White does not have any actual or potential conflict of interest in relation to this CE Activity
UC/HH EPCUC/HH EPC
Learning Objectives By attending this program, participants
should be able to: Identify and describe evidence based
medicine Identify and describe the fundamental
components of a meta-analysis Describe the importance of using meta-
analysis in key areas of clinical practice
UC/HH EPCUC/HH EPC
What is Happening in Healthcare? Increase number and expense of tests and
treatments available (60% of the growth in costs) Monoclonal antibodies, ICDs, erythropoietin
Aging of Baby Boomers Annual prescriptions filled increased by 1.5 billion
over ten years Healthcare costs exploded over 40 years
Cost have grown by 2.5X more than economy annually from 1960-present
Emmanuel EJ. AHA QCOR Conference.
UC/HH EPCUC/HH EPC
A Lot of Money Spent $2.7 trillion spent on healthcare in 2008
1 out of every 6 dollars spent in US How big is a trillion?
1 billion seconds ago Richard Nixon resigned 1 trillion seconds ago was 30,000 years BC
If spending continues to rise at this rate by 2082, 100% of GDP will be spent on healthcare
$200 billion spent on prescription drugs
Emmanuel EJ. AHA QCOR Conference.
UC/HH EPCUC/HH EPC
Evidence-Based Medicine Model: What is it?
“Evidence-based medicine is the conscientious, explicit and judicious use of current best evidence in making decisions about the care of individual patients.” Not simply cookbook medicine but integration of
evidence into practice and knowing when that evidence applies to particular patient and when it does not
Sackett D. BMJ 1996;312:72.
UC/HH EPCUC/HH EPC
Evidence-Based Medicine (EBM) EBM benefits
Maximizes benefits Encourages accountability Enhances efficiency Diminishes harms
EBM implementation Develop consensus on evidence-based practice Disseminate evidence and recommendations to decision-
makers Create incentives to practice EBM
Quality measures Pay-for-performance
Sackett DL. BMJ 1996;312:71–72.; Antman EM. JAMA 1992;268:240-8.; Alexander JA. Health Care Manage Rev 2007;32:150-9.
UC/HH EPCUC/HH EPC
Where Does Evidence Come From? Good evidence:
Both benefits and harms evaluated Evaluated strong endpoints
Strong endpoints: survival, risk of MI, cancer recurrence rate, quality of life, cost-effectiveness
Weak endpoints: blood pressure, cholesterol, glucose
Evaluated effectiveness Effectiveness includes efficacy and applicability
Subpopulations such as women, ethnic minorities, and other groups evaluated
Treadwell JR. BMC 2006;6: doi:10.1186/1471-2288-6-52
UC/HH EPCUC/HH EPC
Which Single Study Should You Use?
Observed Effect+ 40%
Random Error + 5%
Systematic Error+ 5%
True Effect+ 30%
True Effect+ 30%
Observed Effect+ 30%
Systematic Error - 5%
Random Error + 5%
True Effect+ 30%
Observed Effect+ 20%
Random Error- 5%
Systematic Error- 5%
True Effect+ 30%
Observed Effect+ 30%
Systematic Error + 5%
Random Error - 5%
Study 1
Study 2
Study 3
Study 4
UC/HH EPCUC/HH EPC
No, Believe Me, This Single Trial Tells The Whole Truth… No, No, There are No Other
Good Trials…
UC/HH EPCUC/HH EPC
Methodological Rigor in ReviewsIndividual Patient
Data Meta-Analyses
Methodological Rigor HIGHLOW
Qualitative Informal and
subjective methodology
Selective, not comprehensive, literature identification
Rarely report literature selection criteria
Subject to systematic and personal biases
Qualitative Formal methods to
find studies but not to evaluate data
Comprehensive literature inclusion
Apply criteria to select high-quality studies
Describe results in evidence tables
Quantitative Systematic
evidence search with statistical analysis
Pool data from multiple studies to estimate summary statistics with confidence intervals
Quantitative Combine and analyze
patient-level data from primary studies to estimate summary statistics
Infrequently seen because of cost, time, and data considerations
Systematic Review w/ Meta-
AnalysesTraditional, Narrative Reviews
Systematic Reviews w/o
Meta-Analysis
Shea B. BMC Medical Res Methodol 2007;7:10 doi:10.1186/1471-2288-7; Lau J. Lancet 1998;351:123-7.Pai M, National Medical Journal of India 2004;17:86-95.
UC/HH EPCUC/HH EPC
Rationale for Using Systematic Reviews
Provide transparent and objective summary of large amounts of data
Help to cohere conflicting data/results of primary research
Form the basis of policymakers work (e.g., risk assessments, economic analyses)
Identify gaps in knowledge, helping define further avenues for research
UC/HH EPCUC/HH EPC
Meta-Analysis Caution!!!
Healthcare decision-makers need to critically evaluate and understand the value of a given meta-analysis If decision-makers simply accept the pooled result without
exploring the meta-analysis further, they pass all the biases and limitations of the meta-analysis on to their decisions
UC/HH EPCUC/HH EPC
Would You Swim in Here??
UC/HH EPCUC/HH EPC
Meta-Analysis: The Source Matters
Multiple Poor Studies
Meta-Analyzed
UC/HH EPCUC/HH EPC
Publication Bias Publication bias is the tendency of certain
types of trials (such as those with the largest effects) to be published
Publication bias increases the risk that the observed effect might not reflect the true effect May negatively impact consistency, precision, and
magnitude of effect Expanding searches to include additional
languages, citation tracking, hand searching, and grey literature can help identify and possibly minimize publication bias
AHRQ Methods Guide, Finding Evidence, Chapter 5.
UC/HH EPCUC/HH EPC
Publication Bias Example Study: 74 antidepressant studies
registered with the FDA 97% of positive studies published 39% of neutral or negative studies published
11 of 14 published in a way that conveyed the positive results but deemphasized the negative
When only published literature was meta-analyzed, a 32% increase in relative effect size occurred versus the more complete dataset of conducted trials
Turner EH. NEJM 2008;358:252-60.
UC/HH EPCUC/HH EPC
Detecting Publication Bias Funnel Plots
A pictorial representation of each study plotted by its effect size on the horizontal axis and a measure of variance on the vertical axis. If the plot represents an inverted symmetrical funnel, it is said that publication bias is unlikely but publication bias cannot be excluded when any other configuration is shown.
Egger’s weighted regression statistics Simpler to interpret compared to a funnel plot Provides a p-value, if <0.05 indicates publication bias cannot
be ruled out. Begg’s test
A p-value <0.05 indicates publication bias cannot be ruled out.
Begg’s test requires a larger number of studies (>15-20) in order to provide robust results
UC/HH EPCUC/HH EPC
Example of a Funnel Plot
Egger’s p-value =0.51
UC/HH EPCUC/HH EPC
Quantifying the Impact of Publication Bias Duval and Tweedie (2000)
Developed the “Trim and Fill” method Uses funnel plot symmetry to estimate the
number of “missing” studies and the magnitudes of their effects.
Then, re-estimates the overall effect size after imputing potentially “missing” studies into the meta-analysis to determine if the results of the original analysis were replicated.
Duval S, Tweedie R. Biometrics 2000;56:455-63
UC/HH EPCUC/HH EPC
Example of a Trim & Fill Plot
0
2
4
6
8
10
12
14
-0.2 0 0.2 0.4 0.6 0.8
Ln RR
Inve
rse
stan
dard
err
orTrim & fill funnel plot
UC/HH EPCUC/HH EPC
Heterogeneity Between study differences
Some differences between studies is expected due to random variation (chance), but other causes are…
Clinical: Different patient populations, interventions, follow-up times, choice and measurement of outcome
Methodological: Different study designs, quality issues Statistical: Numerical variation in treatment effects
UC/HH EPCUC/HH EPC
How to Assess for Heterogeneity
Summary meta-analysis plot [fixed effects]
combined
Study B
Study A
WMD (95% confidence interval)
Summary meta-analysis plot [fixed effects]
combined
Study B
Study A
WMD (95% confidence interval)
Low (No) Heterogeneity High Heterogeneity
UC/HH EPCUC/HH EPC
How to Assess for HeterogeneityStatistically
Chi-squared test (Cochrane Q statistic) Measures if observed variation is due to chance This test is problematic and has lower power, so if there
are few studies, it may not detect heterogeneity A p-value <0.10 is typically considered significant
I2 (calculated from Q statistic) Gives % of variation likely due to heterogeneity <25% low heterogeneity; 25% - 75% moderate
heterogeneity; >75% high heterogeneity
Higgins et al. BMJ 2003; 327:557-60
UC/HH EPCUC/HH EPC
How to Handle Heterogeneity Use caution when pooling studies that
are not similar clinically, or that have different study designs or where the treatment effects seem inconsistent
If there is a large amount of heterogeneity, explore it… Use subgroup analyses and/or meta-
regression and sensitivity analyses
UC/HH EPCUC/HH EPC
Applicability: PopulationConditions That Limit Applicability
Narrow eligibility criteria, high exclusion rate
Differences between study population and patients in community
Narrow or unrepresentative severity or stage of illness
Run in periods with high exclusion rates.
Events rates markedly different than in community
Disease prevalence in study population different than in community
UC/HH EPCUC/HH EPC
Applicability: InterventionConditions That Limit Applicability
Regimen not reflective of current practice
Intensity of intervention not feasible for routine use
Monitoring practices or visit frequency not used in practice
Versions not in common use
Co-interventions that likely modify effectiveness of therapy
Level of training not widely available
UC/HH EPCUC/HH EPC
Comparator, Outcomes, and Applicability
Conditions That Limit Applicability
ComparatorRegimen not reflective of current practice
Use of substandard alternative therapy
OutcomesSurrogate endpoints, improper definitions for outcomes, composite endpoints
UC/HH EPCUC/HH EPC
Systematic Reviews and U.S. Policymakers Findings from systematic reviews are
being used increasingly by U.S. policymakers Since 1999, the Centers for Medicare
and Medicaid Services (CMS) has commissioned systematic reviews as a step in making national coverage decisions
UC/HH EPCUC/HH EPC
AHRQ Evidence-Based Practice Centers EPCs develop evidence reports and
technology assessments on topics relevant to clinical, social science/behavioral, economic, and other health care organization and delivery issues— Specifically those that are common,
expensive, and/or significant for the Medicare and Medicaid populations
http://www.ahrq.gov/clinic/epc/
UC/HH EPCUC/HH EPC
UCONN/HH EPCDirector:
C. Michael White, Pharm.D., FCP, FCCP UCONN School of Pharmacy
Associate Director/Medical Chief:Jeffrey Kluger, MD, FACC
Hartford Hospital
Supporting Investigators:William Baker, Pharm.D., BCPS
Ripple Talati, Pharm.D., Vanita Tongbram, MBBS, MPH,Ajibade Ashaye, MBBS, MPH,
Wendy Chen, Pharm.D.,Jennifer Colby, Pharm.D.,
Jennifer Scholle, Pharm.D.,Soyon Lee, Pharm.D.
Hartford Hospital/UCONN School of Pharmacy
Medical Librarian:Sharon Giovenale, MS
UCONN School of Pharmacy
Medical Editor:Robert Quercia, MS
Hartford Hospital
Statistician:Jeffrey Mather, MSHartford Hospital
Co-Director/Methods Chief:Craig I. Coleman, Pharm.D.UCONN School of Pharmacy
Project Manager:Diana Sobieraj, Pharm.D.
Hartford Hospital
Content Experts:
Jay Lieberman, MD, Charles Lapin, MD, Raymond McKay, MD, Francis Kiernan, MD, Isaac Silverman, MD, Jennifer Ellis, Pharm.D., Nancy
Rodriguez, PhD, Others to Come
Hartford Hospital, UCHC, CCMC, UCONN
UC/HH EPCUC/HH EPC
Systematic Review Programmatic Themes
Nutraceuticals
Movement Disorders
Cardiology
Internal Medicine
Soluble FibersEchinacea Herbs &
SpicesMagnesium
Early & Late Parkinson’s
Restless Legs
Syndrome
Benefits & Harms of Statins
ACE I and ARBs in
Preserved LV Function
Prevention of Atrial
Fibrillation
rhGH in Cystic
Fibrosis
Asthma/ COPD
ACEI/ARB & DM
Nephro
ICDs
CoQ10 & HF
Warfarin/INR
Control
VTE Ortho Surgery
Guidelines
Transfusion in
Transplant
UC/HH EPCUC/HH EPC
ACE inhibitors or ARBs in CAD ACE inhibitors and ARBs prolong
survival in MI pts with LVD What is the benefit in CAD pts with
preserved LV function? CMS discussing making ACE inhibitor or
ARB use a performance measure Needs CER to discern evidence
UC/HH EPCUC/HH EPC
Mortality and Nonfatal MI
Total Mortality Nonfatal MIRelative risk meta-analysis plot (random effects)
0.2 0.5 1 2 5
TRANSCEND, 2008 1.05 (0.91, 1.20)
PEACE, 2004 0.89 (0.77, 1.03)
CAMELOT, 2004 1.30 (0.47, 3.56)
EUROPA, 2003 0.89 (0.78, 1.02)
SCAT, 2000 0.73 (0.31, 1.74)
PART-2, 2000 0.64 (0.35, 1.17)
HOPE, 2000 0.85 (0.76, 0.95)
combined [random] 0.91 (0.84, 0.98)
relative risk (95% confidence interval)
Relative risk meta-analysis plot (random effects)
0.2 0.5 1 2
PEACE, 2004 1.00 (0.84, 1.20)
CAMELOT, 2004 0.56 (0.27, 1.16)
EUROPA, 2003 0.78 (0.67, 0.90)
SCAT, 2000 0.59 (0.24, 1.42)
PART-2, 2000 0.95 (0.51, 1.76)
HOPE, 2000 0.78 (0.67, 0.91)
combined [random] 0.83 (0.73, 0.94)
relative risk (95% confidence interval)
Ann Intern Med 2009;151:861-71.
UC/HH EPCUC/HH EPC
Stroke and Nonfatal MI
Stroke Composite EndpointRelative risk meta-analysis plot (random effects)
0.01 0.1 0.2 0.5 1 2 5 10
TRANSCEND, 2008 0.83 (0.65, 1.06)
PEACE, 2004 0.77 (0.57, 1.04)
CAMELOT, 2004 0.65 (0.27, 1.53)
EUROPA, 2003 0.96 (0.73, 1.26)
SCAT, 2000 0.22 (0.05, 0.91)
PART-2, 2000 1.76 (0.55, 5.57)
HOPE, 2000 0.69 (0.57, 0.84)
combined [random] 0.79 (0.67, 0.93)
relative risk (95% confidence interval)
Relative risk meta-analysis plot (random effects)
0.5 1 2
TRANSCEND, 2008 0.88 (0.77, 1.00)
PEACE, 2004 0.94 (0.82, 1.07)
HOPE, 2000 0.79 (0.72, 0.87)
combined [random] 0.86 (0.77, 0.95)
relative risk (95% confidence interval)
Ann Intern Med 2009;151:861-71.
UC/HH EPCUC/HH EPC
ACEIs and ARBs in Close Proximity to CABG or PTCA
Relative risk meta-analysis plot (random effects)
0.01 0.1 0.2 0.5 1 2 5 10 100
IMAGINE, 2008 0.99 (0.60, 1.66)
AACHEN, 2006 0.91 (0.05, 15.56)
QUIET, 2001 0.99 (0.59, 1.67)
PARIS, 2001 0.98 (0.06, 16.77)
APRES, 2000 0.25 (0.06, 0.99)
MARCATOR, 2000 2.35 (0.38, 14.63)
combined [random] 0.94 (0.67, 1.34)
relative risk (95% confidence interval)
Relative risk meta-analysis plot (random effects)
0.01 0.1 0.2 0.5 1 2 5 10 100
IMAGINE, 2008 0.76 (0.40, 1.43)
AACHEN, 2006 0.45 (0.06, 3.38)
QUIET, 2001 0.89 (0.58, 1.38)
PARIS, 2001 2.94 (0.25, 35.36)
MARCATOR, 1995 1.13 (0.53, 2.43)
combined [random] 0.89 (0.65, 1.24)
relative risk (95% confidence interval)
Relative risk meta-analysis plot (random effects)
0.01 0.1 0.2 0.5 1 2 5
IMAGINE, 2008 1.07 (0.52, 2.17)
APRES, 2000 0.33 (0.03, 3.95)
combined [random] 1.01 (0.50, 2.04)
relative risk (95% confidence interval)
Total MortalityMI
Stroke
Ann Intern Med 2009;151:861-71.
UC/HH EPCUC/HH EPC
Harms Associated with ACE Inhibitors
Conclusions:
Favorable balance of benefits to harms in most patients, but not for those recently undergoing CABG or PTCA
UC/HH EPCUC/HH EPC
Systematic Review in Pediatrics
rhGH 0.27-0.35mg/kg/wk in CF Trials are rhGH vs. no therapy except
for one placebo controlled trial Sample sizes are small, most
underpowered Perfect role for systematic review
Phung OJ. Pediatrics 2010;21:347-54.
UC/HH EPCUC/HH EPC
rhGH Improves AnthropometricsChange in Height (cm) from Baseline
-1.0 1.5 4.0 6.5 9.00
I2 = 77.3%Egger’s p-value = NA
Weighted Mean Difference (95% Confidence Interval)
Hardin, 2005b
Hutler, 2002
Hardin, 2001
Combined
3.90 (0.52, 7.28)
1.40 (-0.07, 2.87)
4.40 (2.95, 5.85)
3.13 (0.88, 5.38)
Change in Height (cm) from Baseline
-1.0 1.5 4.0 6.5 9.00
I2 = 77.3%Egger’s p-value = NA
Weighted Mean Difference (95% Confidence Interval)
Hardin, 2005b
Hutler, 2002
Hardin, 2001
Combined
3.90 (0.52, 7.28)
1.40 (-0.07, 2.87)
4.40 (2.95, 5.85)
3.13 (0.88, 5.38) Change in Weight (kg) from Baseline
-2.0 0.5 3.0 5.5 8.0 10.5
Stalvey, 2008
Schnabel, 2007B
Schnabel, 2007A
Hardin, 2005b
Hutler, 2002
Hardin, 2001
0
I2 = 49%Egger’s p-value = 0.18
Weighted Mean Difference (95% Confidence Interval)
Combined
1.00 (0.13, 1.87)
1.00 (-0.35, 2.35)
0.80 (-0.78, 2.38)
5.50 (1.76, 9.24)
1.00 (-1.05, 3.05)
2.80 (1.27, 4.33)
1.48 (0.62, 2.33)
Change in Weight (kg) from Baseline
-2.0 0.5 3.0 5.5 8.0 10.5
Stalvey, 2008
Schnabel, 2007B
Schnabel, 2007A
Hardin, 2005b
Hutler, 2002
Hardin, 2001
0
I2 = 49%Egger’s p-value = 0.18
Weighted Mean Difference (95% Confidence Interval)
Combined
1.00 (0.13, 1.87)
1.00 (-0.35, 2.35)
0.80 (-0.78, 2.38)
5.50 (1.76, 9.24)
1.00 (-1.05, 3.05)
2.80 (1.27, 4.33)
1.48 (0.62, 2.33)
Height improves 3.13 cm more with rhGH
Weight improves 1.48 kg more with rhGH
UC/HH EPCUC/HH EPC
rhGH in CF: Pulmonary Function Change in FEV1 (L) from Baseline
-0.5 0.5 1.0 1.5
Hardin, 2006
Hardin, 2005c
Hardin, 2005b
Hutler, 2002
0
I2 = 43.2%Egger’s p-value = 0.11
Weighted Mean Difference (95% Confidence Interval)
Combined
0.20 (-0.01, 0.41)
0.60 (-0.05, 1.25)
0.64 (0.05, 1.23)
0.04 (-0.16, 0.24)
0.23 (0.01, 0.46)
Change in FEV1 (L) from Baseline
-0.5 0.5 1.0 1.5
Hardin, 2006
Hardin, 2005c
Hardin, 2005b
Hutler, 2002
0
I2 = 43.2%Egger’s p-value = 0.11
Weighted Mean Difference (95% Confidence Interval)
Combined
0.20 (-0.01, 0.41)
0.60 (-0.05, 1.25)
0.64 (0.05, 1.23)
0.04 (-0.16, 0.24)
0.23 (0.01, 0.46)Change in FVC (L) from Baseline
0.75 1.50 2.25
Hardin, 2006
Hardin, 2005c
Hardin, 2005b
0
I2 = 55%Egger’s p-value = NA
Weighted Mean Difference (95% Confidence Interval)
Combined
1.00 (0.32, 1.68)
0.90 (0.25, 1.55)
0.40 (0.19, 0.61)
0.67 (0.24, 1.09)
Change in FVC (L) from Baseline
0.75 1.50 2.25
Hardin, 2006
Hardin, 2005c
Hardin, 2005b
0
I2 = 55%Egger’s p-value = NA
Weighted Mean Difference (95% Confidence Interval)
Combined
1.00 (0.32, 1.68)
0.90 (0.25, 1.55)
0.40 (0.19, 0.61)
0.67 (0.24, 1.09)
Absolute FEV1 improves 0.28 L more with rhGH
Absolute FVC improves 0.67 L more with rhGH
UC/HH EPCUC/HH EPC
rhGH in CF: Pulmonary FunctionChange in %FEV1 from Baseline
-20 -10 10 20 30
Schnabel, 2007B
Schnabel, 2007A
Hardin, 2005c
Schibler, 2003
Hardin, 2001
0
I2 = 0%Egger’s p-value = 0.56
Weighted Mean Difference (95% Confidence Interval)
Combined
2.50 (-9.56, 14.56)
3.30 (-9.55, 16.15)
2.00 (-17.21, 21.21)
1.20 (-9.88, 12.28)
4.00 (-10.83, 18.83)
2.43 (-3.99, 8.85)
Change in %FEV1 from Baseline
-20 -10 10 20 30
Schnabel, 2007B
Schnabel, 2007A
Hardin, 2005c
Schibler, 2003
Hardin, 2001
0
I2 = 0%Egger’s p-value = 0.56
Weighted Mean Difference (95% Confidence Interval)
Combined
2.50 (-9.56, 14.56)
3.30 (-9.55, 16.15)
2.00 (-17.21, 21.21)
1.20 (-9.88, 12.28)
4.00 (-10.83, 18.83)
2.43 (-3.99, 8.85)
% Predicted FEV1 nonsignificantly improves 2.43% more with rhGH
UC/HH EPCUC/HH EPC
rhGH: Improves Bone Mineralization
Change in Bone Mineral Content (g) from Baseline
300 600 900
Hardin, 2006
Hardin, 2005c
Hardin, 2005b
Hardin, 2005a
0
I2 = 96.1%Egger’s p-value = 0.82
Weighted Mean Difference (95% Confidence Interval)
Combined
223 (203, 243)
650 (427, 873)
142 (125, 159)
59 (12, 106)
192 (110, 273)
Change in Bone Mineral Content (g) from Baseline
300 600 900
Hardin, 2006
Hardin, 2005c
Hardin, 2005b
Hardin, 2005a
0
I2 = 96.1%Egger’s p-value = 0.82
Weighted Mean Difference (95% Confidence Interval)
Combined
223 (203, 243)
650 (427, 873)
142 (125, 159)
59 (12, 106)
192 (110, 273)
Bone mineral content improves 192g more with rhGH
UC/HH EPCUC/HH EPC
rhGH: HospitalizationsHospitalization Rate (per year) During Therapy
-4 -3 -2 -1
Hardin, 2006
Hardin, 2005c
Hardin, 2005b
Hardin, 2001
0
I2 = 0%Egger’s p-value = 0.98
Weighted Mean Difference (95% Confidence Interval)
Combined
-1.30 (-2.20, -0.40)
-1.81 (-2.38, -1.24)
-1.90 (-3.36, -0.44)
-1.50 (-2.07, -0.93)
-1.62 (-1.98, -1.26)
Hospitalization Rate (per year) During Therapy
-4 -3 -2 -1
Hardin, 2006
Hardin, 2005c
Hardin, 2005b
Hardin, 2001
0
I2 = 0%Egger’s p-value = 0.98
Weighted Mean Difference (95% Confidence Interval)
Combined
-1.30 (-2.20, -0.40)
-1.81 (-2.38, -1.24)
-1.90 (-3.36, -0.44)
-1.50 (-2.07, -0.93)
-1.62 (-1.98, -1.26)
1.62 fewer hospitalizations per year (studies 6-12 mo)
UC/HH EPCUC/HH EPC
rhGH on Final Health Outcomes
Mortality: no data IV antibiotic use: no data Osteopenia/porosis: no data Pneumonia: no data HRQoL: no data
UC/HH EPCUC/HH EPC
rhGH: Harms
Serum glucose increases HbA1c does not change
IGF-I concentrations significantly increase >100ng/mL with rhGH than with control Found to be a marker of neoplasm in past
observational studies Cancer: no data
UC/HH EPCUC/HH EPC
Balance of Benefits to Harms rhGH reduces hospitalizations (SOE:
Moderate) - an important intermediate outcome
rhGh therapy improves height, weight, and pulmonary function (SOE: Moderate) but may or may not impact final health outcomes Epidemiologic links controversial Absolute vs. % Predicted FEV1
rhGH improves BMC (SOE: Low) NaFl supplements improve bone mineral content
but not fractures Does rhGH reduce fractures?
Risk of DM low but possible risk of neoplasm (SOE: Low)
UC/HH EPCUC/HH EPC
Grab Bag Some misc results from our meta-
analyses
UC/HH EPCUC/HH EPC
What a Difference a Y-Chromosome Makes! ICD Survival
0.2 0.5 1 2 5
combined 0.88 (0.63, 1.22)
DEFINITE 1.12 (0.49, 2.62)
MADIT-II 0.58 (0.30, 1.15)
SCD-HeFT 0.96 (0.58, 1.61)
DINAMIT 1.00 (0.49, 2.14)
Hazard Ratio (95% confidence interval)
0.2 0.5 1 2
combined 0.74 (0.60, 0.91)
COMPANION 0.65 (0.45, 0.92)
DEFINITE 0.49 (0.27, 0.90)
MADIT-II 0.70 (0.55, 0.93)
SCD-HeFT 0.73 (0.57, 0.93)
DINAMIT 1.16 (0.78, 1.69)
Hazard Ratio (95% confidence interval)
Men = 36% HR Reduction Significant
Women = 12% HR Reduction, Not Significant
Henyan N. J Intern Med 2006;260:467-73.
UC/HH EPCUC/HH EPC
Combined Cardiovascular Events by Gender with Statins
Summary meta-analysis plot [random effects]
0.2 0.5 1 2
combined 0.76 (0.70, 0.81)
ALLHAT LLT 0.84 (0.71, 1.00)
LIPS 0.79 (0.64, 0.98)
ASCOT LLA 0.36 (0.21, 0.49)
HPS 0.78 (0.74, 0.83)
LIPID 0.76 (0.68, 0.85)
4S 0.66 (0.58, 0.76)
CARE 0.82 (0.72, 0.92)
PROSPER 0.77 (0.65, 0.92)
AFCAPS/TexCAPS 0.77 (0.64, 0.93)
CIS 0.77 (0.41, 1.44)
CCAIT 0.66 (0.31, 1.42)
relative risk (95% confidence interval)
Dale KM. CMRO 2007;23:565-74.
Summary meta-analysis plot [random effects]
0.2 0.5 1 2 5
combined 0.79 (0.69, 0.90)
ALLHAT LLT 1.02 (0.81, 1.28)
LIPS 0.66 (0.38, 1.14)
GREACE 0.46 (0.27, 0.72)
HPS 0.81 (0.72, 0.92)
PROSPER 0.96 (0.79, 1.18)
LIPID 0.87 (0.67, 1.13)
4S 0.65 (0.47, 0.91)
CARE 0.58 (0.42, 0.81)
AFCAPS/TexCAPS 0.67 (0.34, 1.31)
PLAC I 1.09 (0.34, 3.52)
CCAIT 1.07 (0.23, 4.88)
relative risk (95% confidence interval)
Men = 30% RR Reduction Women 30% RR Reduction
UC/HH EPCUC/HH EPC
Any Questions?