• PubMed was searched for randomized clinical trials published between 2007 and 2012 in the New England Journal of Medicine (NEJM)

• Trials were included if they enrolled at least 5,000 patients and had a

cardiovascular primary endpoint

• Rates of LTFU and WDC were determined from the primary manuscripts and online supplementary material when necessary

Robert W. Harrison, MD; Daniel Wojdyla, MS; Kenneth W. Mahaffey, MDDuke Clinical Research Institute, Durham, NC

Conclusions

Robert W. Harrison, MDDuke Clinical Research Institute2400 Pratt St.Durham, NC 27705Robert.w.harrison@duke.edu

Lost to follow-up in contemporary global cardiovascular randomized clinical trials

Background

Objectives

Methods

Results

Trial Acronym Year of Publicat

ion

Median Follow-up

(days)

Number of Subjects

LTFU

N (%)

WDC

N (%)

ACCOMPLISH 2008 1086 11,506 117 (1.02) 5 (0.04)

ACCORD-Lipid 2010 1716 5,518 56 (1.00) -

ACTIVE-A 2009 1314 7,554 43 (0.57) -

APPRAISE-2 2011 241 7,392 50 (0.68) 81 (1.10)

ARISTOTLE 2011 657 18,201 69 (0.38) 199 (1.10)

ASCEND-HF 2011 30 7,141 8 (0.11) 30 (0.42)

ATLAS ACS 2 2012 398 15,526 37 (0.24) 1271 (8.30)

AVERROES 2011 402 5,599 - -

CHAMPION-PCI 2009 30 8,877 69 (0.78) 2 (0.02)

CHAMPION-PLATFORM 2009 30 5,362 19 (0.35) 6 (0.11)

CURRENT OASIS-7 2010 30 25,086 25 (0.10) -

EARLY-ACS 2009 30 9,492 11 (0.12) 14 (0.15)

ILLUMINATE 2007 550 15,067 39 (0.26) 283 (1.88)

JUPITER 2008 694 17,802 81 (0.46) -

NAVIGATOR 2010 2373 9,306 905 (9.72) 306 (3.29)

ONTARGET 2008 1703 25,620 43 (0.17) -

ORIGIN 2012 2263 12,536 18 (0.14) 32 (0.26)

PLATO 2009 277 18,624 5 (0.03) -

PRoFESS 2008 913 20,332 125 (0.61) -

RE-LY 2009 730 18,113 20 (0.11) -

ROCKET AF 2011 707 14,264 32 (0.22) 447 (3.10)

TRA-2P 2012 912 26,449 32 (0.12) 532 (2.00)

TRACER 2012 502 12,944 15 (0.12) 581 (4.50)

TRILOGY-ACS 2012 517 9,326 7 (0.08) 19 (0.20)

TRITON-TIMI 38 2007 441 13,608 14 (0.10) -

Median (IQR)550

(277 - 913)

12,944

(8877 - 18113)

0.23%

(0.12 - 0.58)

1.10%

(0.18 – 2.55)

Table 1: Study Characteristics• Subjects in clinical trials may

choose to withdrawal their consent (WDC) to participate, or become lost to follow-up (LTFU) prior to study completion

• High rates of LTFU may introduce uncertainty around the validity of the results of clinical trials

• Incomplete follow-up data may lead to increased scrutiny on behalf of regulatory agencies, and may threaten approval of investigational products 1

• Currently, there is no standard method for reporting LTFU and WDC

• Review the methods for reporting LTFU and WDC in large contemporary global cardiovascular clinical trials

• Determine the rates of LTFU and WDC in large contemporary global cardiovascular clinical trials

Disclosures:Harrison, RW: None

Wojdyla, D: None

Mahaffey, KW: Consulting fees/honoraria: Johnson&Johnson, AstraZeneca, Exeter Group, Orexigan, Biotronik, Amgen, Genentech, Sun Pharma, Forest, Adolor, WebMD, Ortho/McNeill, Haemonetics, Daiichi Sankyo, Pfizer, South East Area Health Education Center, Elsevier (AHJ), Eli Lilly, Gilead Science, Bristol Myers Squibb, Glaxo Smith Kline, Novartis Pharmaceutical, Medtronic, Merck, Sanofi-Aventis, Boehringer Ingleheim, Bayer, Polymedix; Research Grants: INC Research, Eli Lilly, Medtronic, Merck , Pozen, Baxter, Cordis, Boehringer Ingelheim, Luitpold, AstraZeneca, Edwards Lifesciences, Bristol Myers Squibb, Abbott Vascular, Portola, Daiichi Sankyo, Sanofi, Guidant , Bayer, Amgen, Glaxo Smith Kline, Johnson & Johnson, Roche Diagnostic, Novartis Pharmaceutical, Amylin, Schering Plough Research Institute, Ikaria, The Medicines Company, Regado, Springer Publishing, KAI

Contact

Reference: 1 FDA Briefing Document for the Cardiovascular and Renal Drugs Advisory Committee (CRDAC). http://www.fda.gov/downloads/AdvisoryCommittees/CommitteesMeetingMaterials/Drugs/CardiovascularandRenalDrugsAdvisoryCommittee/UCM304755.pdf. May 23, 2012; Accessed March 1, 2013

• There is great variability in the manner in which LTFU and WDC are reported

• 40% of studies did not contain a CONSORT diagram, and 40% of studies did not describe LTFU or WDC by treatment arm

• Overall, the proportions of LTFU were low, but in some

trials over 100 patients had LTFU

• WDC occurred more frequently but was only reported in 60% of the trials

• These results emphasize the need to standardize reporting of LTFU and WDC as important trial metrics of quality and to develop strategies to minimize their occurrence

Number %

Reported data on LTFU 24/25 96%

Contain a CONSORT diagram 15/25 60%

Separately report LTFU by treatment arm 16/25 64%

Separately report WDC and LTFU 15/25 60%

Table 2: Reporting of LTFU and Withdrawn Consent

PLATO

CURRENT OASIS-7

TRITON-TIMI 38

RE-LY

ONTARGET

EARLY-ACS

TRILOGY-ACS

ORIGIN

JUPITER

CHAMPION-PLATFORM

ASCEND-HF

ACTIVE-A

PRoFESS

CHAMPION-PCI

ACCORD-Lipid

ACCOMPLISH

ARISTOTLE

APPRAISE-2

TRA-2P

ILLUMINATE

ROCKET AF

TRACER

ATLAS

NAVIGATOR

AVERROES

0% 2% 4% 6% 8% 10% 12% 14%

Proportion

Not Reported

Figure: Proportion of subjects with LTFU or WDC

LTFU Withdrawn Consent

Median (IQR)

0.23%(0.12%-0.58%)

1.10%(0.18%-2.55%)

Range 0.03% - 9.7%

0.02% - 8.3%

Fold variation

323x 415x

WDC

LTFU