1. How to do a study?
Prof. P. Devroey

2. Why to do a study? To know more To improve practice
To improve the results
To treat better
To reduce the cost

3. It depends on what you are seeking for..
What kind of study should we perform?
It depends on what you are seeking for..

4. Type of studies and level of evidence
RCT
Cohort studies
Case series case reports
LEVEL I evidence
LEVEL II evidence
LEVEL III evidence

5. THE STUDY PROTOCOL …..possibly the most important part of the trial

6. The proper study protocol for RCT
Inclusion-exclusion criteria
Appropriate statistical analysis
Randomization procedure
Clearly defined interventions
One primary endpoint
Data management

7. Ethical considerations
Internal review board approval
Ethics committee
Local or national
Informed consent
Trial registration (number)
i.e.

8. Institutional Review Board (IRB) approval
All institutions should have an IRB to evaluate whether it is ethical to conduct the study
Submit the protocol and wait for acceptance prior to conducting a trial
Explain in detail the rationale, the population, the interventions and the goals of the study

9. Register your trial
After approval of IRB and by the ethics committee register the trial (e.g. clinicaltrials.gov)
Journals require trial registration prior to conduction of the study

10. The research question One question and one answer
In accordance with the available evidence
Why is such a study valuable?
Can it change clinical practice?

11. The population to include..
Clearly defined
Easy to recruit
Easy to follow

12. Unclear population…not replicable results
47 randomized trials using 41 definitions for poor ovarian responders
No more than 3 trials used the same definition
Even trials from the same research group used different definition
Who are the poor ovarian
responders?
Polyzos and Devroey. Fertil Steril 2011

13. Sample size calculation
Power-sample calculation
80% power, level of significance 0.05
Estimate your sample based on previous evidence
Do not make your sample size based on unrealistic assumptions
During the protocol formulation describe the statistics you will use

14. The feasibility of conducting the study
Resources available
Available number of patients
Easy to follow your patients

15. Select your people for conducting the trial
Investigators
Study nurse
Co-ordinator
Statistician
Follow rigorously the ethical guidelines
Do not attempt to publish at any cost!
Define the mechanism of publishing: primary author and co-authors

16. OCP + GnRH antagonist ( n) (%)
Oral contraceptive pills in GnRH antagonist protocol versus long protocol
OCP + GnRH antagonist ( n) (%)
Long Protocol (n) (%)
Ongoing PR
55/115 (48)
61/113 (54)
Multiples
15/55 (27)
18/61 (30)
Implantation Rate
75/207 (36)
80/204 (39)
Live birth rate
51/115 (44)
53/113 (47)
Garcia-Velasquo FS 2011

17. Results after 1 embryo SET GnRH antagonist
Standard FSH
Low dose hCG
Live birth rate
Cycle (%)
10 / 35 (29)
13 / 35 (37)
OPU (%)
10 / 32 (31)
13 / 29 (45)
ET (%)
10 / 29 (35)
13 / 27 (48)
Blockeel C et al HR 2009

18. GnRH agonist GnRH antagonist Standard Low dose P FSH (U) 2800 1900
< 0.001
1617
1273
0.001
Duration (d)
11.6
11.9
8.2
8.7
rFSH (U)
8.6
6.4
E2 (ng/ml)
2358
3235
< 0.05
2044
2200 NS
Prog (ng/ml)
1.1
1.2
Filicori FS Blockeel HR 2009

19. Cost evaluation GnRH agonist long protocol
n = 1000 cycles
GnRH agonist long protocol
GnRH antagonist + low dose hCG
Cost (Euro)
Price difference (euro)
SAVINGS
Personal communication

20. Randomization Patients received 10.000 IU of hCG
as soon as ≥ 3 follicles of ≥ 17 mm were present in ultrasound
early hCG group, patients or
2 days later after this criterion was met
late hCG group, patients
Kolibianakis Albano Camus Tournaye Van Steirteghem Devroey FS 2004

21. Prolongation of the follicular phase in IVF results in a lower probability of pregnancy
Early-hCG group
Late-hCG group P
Ongoing pregnancy rate per OPU (n)
35.6%
(69/194)
25%
(49/196)
0.027
Ongoing pregnancy rate per ET (n)
39.2%
(69/176)
27.7%
(49/177)
0.024
Ongoing implantation rate (n)
22.6%
(87/385)
15.1%
(58/383)
0.009
Kolibianakis FS 2004

22. Elevated progesterone levels at initiation of stimulation are associated with a significantly lower chance of pregnancy
Normal P group
High P group P
Difference (95% CI)
Ongoing pregnancy rate
Per started cycle % (n)
31.8 (124/390)
5.0 (1/20)
0.011
26.8 ( )
Per oocyte retrieval % (n)
33.8 (124/367)
6.3 (1/16)
0.026
27.5 ( )
Further research Initiation of antagonist on day 1?
Kolibianakis HR 2004

23. Impact on cycle outcome
Bosch et al HR 2010

24. Endometrial biopsy on the day of ovulation , natural cycle
No secretory features

25. Endometrial biopsy on the day of oocyte retrieval , GnRH agonist and gonadotrophin stimulation cycle
Clear secretory features

26. Is there any relation between endometrial advancement and ongoing pregnancy rates ?
YES or NO
Answer : Yes
≤ 3 days
> 3 days P
hMG / agonist
10 / 32
0 / 7
recFSH / antagonist
8 / 49
0 / 6
TOTAL
18 / 81
0 / 13
< 0.05
Endometrial advancement
Kolibianakis FS 2002

27. Recombinant LH after antagonist initiation
Pill pre-treatment/ 3 day interval, variable starting dose of rec FSH
Single dose antagonist administration by a follicle of 14-16mm
Cedrin-Durnerin HR 2004

28. Well-written papers are Read Remembered Cited
31 March 2011
Well-written papers are
Read
Remembered
Cited
Poorly written papers are not…
Agonist triggering and freeze-all: How and when to freeze - D. Stoop

29. Which journal? Select the journal relevant to the work done
31 March 2011
Which journal?
Select the journal relevant to the work done
Read the ‘guidelines for authors’ on the website of the journal
Depends on the ‘quality’ of the research performed
Innovative
Design of the study
Size of the study
Human or animal
Effect on clinical practice
Agonist triggering and freeze-all: How and when to freeze - D. Stoop

30. Hirsch factor Number of citations by other papers
P. Devroey
Citations:
h-index: 109

31. Conclusion The research needs to be innovative
A randomized controlled trial is preferential
Sample size and power calculation have to be respected
Any prospective trial needs registration
Only one primary endpoint
A positive or a negative finding of the research question is of equal value
Acknowledgements to Helena Deryckere