Upload
ronald-bruce
View
213
Download
0
Tags:
Embed Size (px)
Citation preview
Analysis of Overall Impact Scoring Trends
within AHRQ Peer Review Study Sections• Gabrielle Quiggle, MPH; Rebecca Trocki, MSHAI; • Kishena Wadhwani, PhD, MPH; Francis Chesley, MD• Office of Extramural Research, Education, and Priority
Populations, Agency for Healthcare Research and Quality
Background: Peer Review
The Agency for Healthcare Research and Quality (AHRQ) has a chartered health services research Initial Review Group (IRG) responsible for the peer review of grant applications submitted for funding opportunity. This IRG is comprised of five subcommittees or study sections:
o Healthcare Systems & Value Research (HSVR)o Healthcare Safety & Quality Improvement Research
(HSQR)o Healthcare Information Technology Research (HITR)o Health Care Effectiveness and Outcomes Research
(HEOR)o Health Care Research Training (HCRT)
Background: Peer Review
Research grant applications submitted to AHRQ are reviewed by one of five standing study section committees.
Applications are submitted in response to a Program Announcement (PA) or Request for Application (RFA).
General research grant mechanisms of interest:R01 – Research project grants – (Independent)R03 – Small research project grantsR18 – Research demonstration and dissemination
project grants
AHRQ uses a 9-point overall impact score system to evaluate the scientific/technical merit of research grant applications submitted to AHRQ for funding opportunities.
Score Descriptor Additional Guidance on Strengths/Weaknesses1 Exceptional Exceptionally strong with essentially no weaknesses2 Outstanding Extremely strong with negligible weaknesses3 Excellent Very strong with only some minor weaknesses4 Very Good Strong but with numerous minor weaknesses5 Good Strong but with at least one moderate weakness6 Satisfactory Some strengths but also some moderate weaknesses7 Fair Some strengths but with at least one major weakness8 Marginal A few strengths and a few major weaknesses9 Poor Very few strengths and numerous major weaknesses
High
Medium
Low
AHRQ Peer Review Scoring Descriptor Table
Non-numeric score options:DF = Deferred, AB = Abstention, CF = Conflict, NP = Not Present, ND = Not Discussed
Background: Scoring
The final overall impact score reflects the average of the impact scores provided by the study section members as a whole (x10). Percentiles are calculated to rank applications relative to each other.
Background: Scoring
10 90High impact Low impact
Final Overall Impact Score
Preliminary Score
1 9
Exceptional Poor
1. To determine whether trends exist in the scoring of research grant applications submitted for funding by AHRQ
2. To assess potential differences in scoring trends between the five (5) AHRQ study sections
Study Objectives
Final impact scores were obtained from the following applications:
o First-time applicationso Received from October 2009 to June 2014 (15
review cycles)o Submitted to one of five AHRQ study sectionso Not withdrawn
Resubmitted applications and applications not discussed (ND) were excluded from sample
Data collection using NIH eRA Commons and NIH Query View & Report Database (QVR system)
Methods: Data Collection
• Means (SD) and medians (range) were calculated for each quarterly review meeting and fiscal year.
• Score trends were assessed by council meeting for each study section, using all application mechanisms.o Subgroup analysis was conducted on applications
considered under general research mechanisms R01, R03, and R18 in FY 2011-2014
• Percentile standardized scores were used to compare score trends between study sections.
Descriptive statistics and linear regression conducted using MS Excel and SAS 9.3.
Methods: Analysis
• AHRQ received 3,370 applications between Fiscal Year 2010 and 2014.
• 1,752 (52%) applications were discussed and received a final overall impact score.
• Slight trends towards lower (better) median scores were found in four out of five AHRQ study sections:
Results
Study section Trend line R2
HEOR -0.67x + 47 0.203
HSQR -0.98x + 50 0.233
HSVR +0.45x + 36 0.110
HITR -0.23x + 35 0.081
HCRT -0.41x + 37 0.107
Results
2010-10 2011-01 2011-05 2011-10 2012-01 2012-05 2012-10 2013-01 2013-05 2013-10 2014-01 2014-0510
20
30
40
50
60
70
80
90
f(x) = − 0.671328671328671 x + 47.1136363636364R² = 0.202824712659489
HEOR (FY 2011 - 2014)
AHRQ Review Cycle
Impact
Score
Scored = 42 ND = 39% ND = 48%
Scored = 45 ND = 13% ND = 22%
Scored = 27 ND = 26% ND = 49%
Scored = 32 ND = 12% ND = 27%
Scored = 30 ND = 24% ND = 44%
Scored = 21 ND = 15% ND = 42%
Scored = 21 ND = 11% ND = 34%
Scored = 20 ND = 5% ND = 20%
Scored = 22 ND = 19% ND = 46%
Scored = 30 ND = 32% ND = 52%
Scored = 13 ND = 8% ND = 38%
Scored = 25 ND = 15% ND = 38%
Results
2011-10 2012-01 2012-05 2012-10 2013-01 2013-05 2013-10 2014-01 2014-0510
20
30
40
50
60
70
f(x) = − 0.983333333333333 x + 49.5833333333333R² = 0.232531730126921
HSQR (FY 2011 - 2014)
AHRQ Review Cycle
Impact
Score
Scored = 31 ND = 25% ND = 45%
Scored = 35 ND = 34% ND = 49%
Scored = 25 ND = 27% ND = 52%
Scored = 21 ND = 22% ND = 51%
Scored = 15 ND = 7% ND = 32%
Scored = 28 ND = 13% ND = 32%
Scored = 27 ND = 29% ND = 52%
Scored = 22 ND = 19% ND = 46%
Scored = 23 ND = 27% ND = 54%
Results
2010-10 2011-01 2011-05 2011-10 2012-01 2012-05 2012-10 2013-01 2013-05 2013-10 2014-01 2014-0510
15
20
25
30
35
40
45
50
55
60
f(x) = 0.445804195804196 x + 35.6439393939394R² = 0.109632792667431
HSVR (FY 2011 - 2014)
AHRQ Review Cycle
Impact
Score
Scored = 35 ND = 36% ND = 51%
Scored = 30 ND = 27% ND = 47%
Scored = 27 ND = 36% ND = 57%
Scored = 32 ND = 27% ND = 46%
Scored = 27 ND = 16% ND = 37%
Scored = 10 ND = 20% ND = 67%
Scored = 14 ND = 11% ND = 44%
Scored = 9 ND = 9% ND = 50%
Scored = 11 ND = 9% ND = 45%
Scored = 26 ND = 18% ND = 41%
Scored = 14 ND = 9% ND = 39%
Scored = 12 ND = 9% ND = 43%
Results
2010-10 2011-01 2011-05 2011-10 2012-01 2012-05 2012-10 2013-01 2013-05 2013-10 2014-01 2014-0510
20
30
40
50
60
70
80
90
f(x) = − 0.234265734265734 x + 35.1060606060606R² = 0.0809757740110276
HITR (FY 2011 - 2014)
AHRQ Funding Cycle
Impact
Score
Scored = 48 ND = 50% ND = 51%
Scored = 35 ND = 29% ND = 45%
Scored = 44 ND = 28% ND = 39%
Scored = 32 ND = 27% ND = 46%
Scored = 29 ND = 31% ND = 52%
Scored = 28 ND = 17% ND = 38%
Scored = 28 ND = 25% ND = 47%
Scored = 17 ND = 14% ND = 45%
Scored = 21 ND = 21% ND = 50%
Scored = 44 ND = 41% ND = 48%
Scored = 26 ND = 24% ND = 48%
Scored = 32 ND = 24% ND = 43%
Results
2010-10 2011-01 2011-05 2011-10 2012-01 2012-05 2012-10 2013-01 2013-05 2013-10 2014-01 2014-0510
20
30
40
50
60
70
80
90
f(x) = − 0.409090909090909 x + 36.9924242424242R² = 0.107237422771403
HCRT (FY 2011 - 2014)
AHRQ Review Cycle
Impact
Score
Scored = 26 ND = 21% ND = 45%
Scored = 36 ND = 17% ND = 32%
Scored = 22 ND = 29% ND = 57%
Scored = 27 ND = 14% ND = 34%
Scored = 24 ND = 8% ND = 25%
Scored = 14 ND = 8% ND = 36%
Scored = 18 ND = 7% ND = 28%
Scored = 17 ND = 16% ND = 48%
Scored = 11 ND = 13% ND = 54%
Scored = 8 ND = 13% ND = 62%
Scored = 20 ND = 11% ND = 35%
Scored = 26 ND = 21% ND = 45%
• Subgroup analysis included 1,086 applications (57% discussed) considered under general research mechanisms R01, R03, and R18.
• Triaging of applications was highamong R03 (54.8%) and R18 (57.5%) applications across all study sections, compared to R01 applications (11.4%).
Mean scores:R01 = 34.2 ± 13.3 to 42.8 ± 12.5R03 = 36.4 ± 15.2 to 41.9 ± 12.8R18 = 38.0 ± 13.6 to 43.1 ± 18.0
Results
159 167
37103
177 203
86
154
H Q E R H SV R H SQ R H T D S
SCORED VS. UNSCORED APPLICATIONS
UNSCORED SCORED
ResultsComparison of score distribution between study sections:
Percentile scores did not differ by study section, adjusting for FY, for R01 (F=0.74, p=0.53), R03 (F=0.31, p=0.82), and R18 (F=0.22, p=0.88).
The analysis of impact scores among study sections as a function of time revealed no statistically significant differences.
AHRQ study sections perform consistently over time, reflecting both the assessments of the reviewers and the quality of the applications.
These results show that careful selection of subject-matter experts, and consistency and uniformity in conducting the evaluation of research grant applications, are the best practices for peer review.
Conclusions
FIRST AUTHOR:• Gabrielle Quiggle, MPH
AHRQ/OEREP staff:• Francis Chesley, MD – Director of OEREP/AHRQ• Kishena Wadhwani, PhD, MPH – Director of Division
of Scientific Review (DSR)/OEREP• Rebecca Trocki, MSHAI – Program Analyst,
DSR/OEREP
Acknowledgements
AHRQ study section review committees: www.ahrq.gov/fund/peerrev/peerdesc.htm www.ahrq.gov/funding/process/study-section/peerdesc.html
AHRQ research announcements:www.ahrq.gov/funding/research/announcements/index.html
AHRQ scoring criteria:www.grants.nih.gov/grants/guide/notice-files/NOT-HS-10-002.html
Resources