Background
Methods
Material
Clinical studies: inclusion criteria
Stakeholders
Methods
Clinical studies
Cost estimation
Stakeholders’ satisfaction
Statistical analysis
Clinical studies
Stakeholders’ satisfaction
Results
pCRFs n = 16 | eCRFs n = 11 | All n = 27 | ||
---|---|---|---|---|
Design
|
Non-interventional studies
| 1 (6%) | 4 (36.4%) | 5 (18.5%) |
Clinical trials
| 15 (94%) | 7 (63.6%) | 22 (81.5%) | |
Randomized
| 10 (63%) | 7 (64%) | 17 (63%) | |
Multicenter
| 10 (63%) | 9 (82%) | 19 (70%) | |
Geographic level
|
International/national
| 7 (44%) | 8 (73%) | 15 (56%) |
Regional
| 9 (56%) | 3 (27%) | 12 (44%) | |
Purpose*
|
Diagnostic
| 1 (6%) | 3 (27%) | 4 (15%) |
Observational
| 0 (0%) | 2 (18%) | 2 (7%) | |
Therapeutic
| 15 (94%) | 6 (55%) | 21 (78%) | |
Risk level
α
|
A
| 2 (13%) | 4 (36.4%) | 6 (22%) |
B
| 4 (25%) | 3 (27%) | 7 (26%) | |
C
| 3 (19%) | 3 (27%) | 6 (22%) | |
D
| 7 (44%) | 1 (9%) | 8 (30%) | |
Clinical trial phase
|
1
| 1 (7%) | 0 (0%) | 1 (4%) |
2
| 4 (27%) | 1 (14%) | 5 (23%) | |
3
| 7 (47%) | 6 (86%) | 13 (59%) | |
4
| 3 (20%) | 0 (0%) | 3 (14%) | |
Median number of patients included*
| 60 | 355 | 80 | |
Q1 = 27 | Q1 = 78 | Q1 = 50 | ||
Q3 = 141 | Q3 = 700 | Q3 = 500 | ||
Median number of centers
| 5 | 10 | 7 | |
Q1 = 1 | Q1 = 6 | Q1 = 1 | ||
Q3 = 10 | Q3 = 13 | Q3 = 12 | ||
Median planned duration of study (months)
| 24 | 27 | 25 | |
Q1 = 16 | Q1 = 18 | Q1 = 16 | ||
Q3 = 40 | Q3 = 36 | Q3 = 36 | ||
Median planned patient follow up (days)
| 137 | 60 | 91 | |
Q1 = 67 | Q1 = 12 | Q1 = 30 | ||
Q3 = 365 | Q3 = 112 | Q3 = 213 | ||
Median number of variables in CRF
| 1,062 | 396 | 1011 | |
Q1 = 669 | Q1 = 153 | Q1 = 286 | ||
Q3 = 1,118 | Q3 = 1,567 | Q3 = 1,126 | ||
Median number of variables in database
β
| 65,928 | 304,929 | 76,692 | |
Q1 = 18,764 | Q1 = 35,250 | Q1 = 20,088 | ||
Q3 = 171,646 | Q3 = 625,865 | Q3 = 304,929 | ||
Median number of full pages in CRF*
| 39 | 17 | 31 | |
Q1 = 28 | Q1 = 9 | Q1 = 17 | ||
Q3 = 44 | Q3 = 30 | Q3 = 44 |
OR | IC | p | ||
---|---|---|---|---|
Geographic level
|
International/national
| 1 | - | 0.14 |
Regional
| 0.292 | 0.056 – 1.525 | ||
Risk level
|
A
| 1 | - | 0.28 |
B
| 0.375 | 0.039 – 3.605 | ||
C
| 0.500 | 0.049 – 5.154 | ||
D
| 0.071 | 0.005 – 1.059 | ||
Number of patients included
| 1.004 | 1.000 – 1.009 | 0.04 | |
Number of centers
| 1.004 | 0.963 – 1.047 | 0.85 | |
Number of variables
| 0.999 | 0.998 – 1.000 | 0.23 | |
Planned duration of study
| 0.987 | 0.928 – 1.050 | 0.68 | |
Number of medical teams involved
| 0.936 | 0.369 – 2.374 | 0.89 |
Clinical studies
Parameter estimate | SE |
Pvalue | ||
---|---|---|---|---|
Duration of the study:
| ||||
CRF*
|
Paper
| - | - | 0.045 |
Electronic
| −10.14 | 4.79 | ||
Participating centers*
| 0.48 | 0.13 | 0.001 | |
Cost of the study (log):
| ||||
CRF
|
Paper
| - | - | 0.41 |
Electronic
| 0.29 | 0.35 | ||
Design*
|
Trial without randomization
| - | - | 0.002 |
Trial with randomization
| −1.33 | 0.40 | ||
Non-interventional study
| −2.51 | 0.62 | ||
Number of patients included*
| −0.001 | 0,0004 | 0.021 |
Stakeholders’ satisfaction
Investigators | CRAs | DMs | ||
---|---|---|---|---|
Respondent
| 34 | 41 | 17 | |
Age
|
< 30
| 0 (0%) | 17 (42%) | 7 (44%) |
30 to 40
| 8 (24%) | 19 (46%) | 7 (44%) | |
> 40
| 26 (76%) | 5 (12%) | 2 (12%) | |
Gender
|
M
| 20 (59%) | 6 (15%) | 10 (59%) |
F
| 14 (41%) | 35 (85%) | 7 (41%) | |
Computer proficiency level
|
Beginner
| 1 (3%) | - | - |
Average
| 19 (56%) | - | - | |
Good
| 14 (41%) | - | - | |
Clinical research experience
|
< 1 year
| - | 4 (10%) | 2 (12%) |
1 to 3 years
| - | 19 (46%) | 7 (41%) | |
3 to 5 years
| - | 11 (27%) | 3 (18%) | |
> 5 years
| - | 7 (17%) | 5 (29%) |
Main features answered | Details | |
---|---|---|
Investigators
| Quality interface (x23) | Fast, simple, without bugs and blocking, with flexible data entry |
Reliable data collection, with few queries (x8) | Alarms and mandatory fields | |
Electronic format (x3) | Allowing data sharing without data recapture and duplication | |
CRAs
| Quality interface (x24) | Fast, simple, without bugs and blocking, with flexible data entry |
Accessible form (x7) | ||
Efficient monitoring (x5) | Real-time queries and remote consultations | |
Quality of form (x5) | No free answers, short and clear questions and few variables | |
Immediate controls during data entry (x3) | ||
Motivation and availability of investigators (x3) | ||
DMs
| Reliable data collection, few queries (x17) | Controls and constraints during data entry and queries emailed automatically |
Quality of database, easy to operate (x5) | ||
Quality of form (x5) | Real-time queries and remote consultations | |
Quality of form (x5) | No free answers, short and clear questions and few variables | |
Quality interface (x3) | Ergonomic | |
Maximum access fees and free action for data managers (x2) |
Investigators | Complaints | About redundancy of data “It doesn’t matter whether it’s on paper or electronic, as long as data are entered only once." |
That some promoters want them to complete a pCRF first as source document and then to re-enter all the data in the eCRF. | ||
About software design companies “By trying to make money, firms that sell this type of CRF software developed templates that do not fit well with the variability of studies and data.” | ||
Hopes | To have their needs taken into account "eCRFs are developed by those who use data but never by those who enter the data and who have, in the present context, less and less availability.“ | |
Working with transportable computers “The graphic tablet, an eCRF transportable to the bedside, is the solution for future. It’s already used in anesthesia with great success.“ | ||
CRAs | Complaints | CTAs needed on site: “Whatever the collection method, investigators don’t have the time …” |
eCRF storage: CRFs still needs to be kept on paper, as source data and to be signed … | ||
DMs | Complaints about CleanWEB consistency management | “Only simple checks can be defined in CleanWEB; more complicated ones must be programmed in SAS after database export.” |
Moreover, “the computer code managing the automatic controls should be accessible and easily understood." | ||
“The database structure isn’t known when designing eCRFs for CleanWEB, yet it is the first thing that must be established. And it’s currently impossible to have a structure that complies with CDISC." |