Skip to main content
main-content

01.12.2008 | Research article | Ausgabe 1/2008 Open Access

BMC Health Services Research 1/2008

Opinions on registering trial details: a survey of academic researchers

Zeitschrift:
BMC Health Services Research > Ausgabe 1/2008
Autoren:
Martin Scherer, Sven Trelle
Wichtige Hinweise

Electronic supplementary material

The online version of this article (doi:10.​1186/​1472-6963-8-18) contains supplementary material, which is available to authorized users.

Competing interests

ST was a member of the Ottawa Group. MS declares that he has no competing interests.

Authors' contributions

MS and ST equally participated in study design, statistical analysis and writing of the manuscript. Both authors had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Both authors read and approved the final manuscript.

Background

Since 2005 the International Committee of Medical Journal Editors (ICMJE) has required prospective registration of clinical studies in order to be considered for publication in one of their journals [1]. Moreover, the World Health Organization (WHO) has established a set of items related to study design and administrative information that should form the minimum set of data in a study register [2]. Several stakeholders were consulted on this WHO Trial Registration Data Set including academic researchers, industry and patient representatives. However, several items were highly controversial and industry representatives demanded that some items should not be disclosed during study conduct [3]. Industry representatives argued that some trial information is sensitive for competitive reason and asked for delayed disclosure of some items e.g. the scientific title of a study, primary and secondary outcomes. However, competition is also of increasing importance in academic research [4].
In 2005 the Ottawa Group published principles of trial registration [5] and a minimum data set for registration is currently being developed [6]. This set requires even more details about a study than the WHO Trial Registration Data Set (see Table 1). However, analyses of trial registration in ClinicalTrials.gov showed heterogeneous quality and completeness of the registration entries indicating varying degrees of comfort with different levels of disclosure [7]. Since nothing is known about the attitudes of academic researchers towards prospective study registration, it is unclear if they would comply with an extension of the requested study information. We therefore surveyed academic researchers about their opinions regarding the registration of study details proposed by the Ottawa Group.
Table 1
Study details required by WHO and Ottawa Group
WHO Trial Registration Data Set
Ottawa Group Data Set (in addition to WHO items)
Trial registration date
Trial acronym
Ethics approval
Trial website
Funding source(s)
Lay description
Primary sponsor
Registration date
Secondary sponsor(s)
Date ethical approval
Coordinating/principle investigator
Date recruitment end
Contact person
Date end of follow-up
Coordinating center
Date trial stop
Recruitment center locations
Trial extensions
Official scientific title
Date primary analyses
Lay title
Name of Research ethics board/institutional
Date recruitment start
Review board (REB/IRB)
Recruitment status
REB trial approval number
Inclusion criteria
Rationale of the trial
Exclusion criteria
References to systematic reviews
Disease/condition
Justification of interventions
Interventions
Trial objectives
Primary endpoint
Study design
Secondary endpoints
Number of arms
Trial phase/study type
Generation of allocation sequence
Target sample size
Randomization
 
Allocation concealment
 
Blinding status
 
Other design features
 
Framework of the study
 
Sample size calculation
 
Planned subgroup analyses
 
Planned analyses methods
 
Consent forms
 
Full protocol
 
Contracts and financial arrangements

Methods

Study design

This study was a survey of academic researchers currently running investigator-initiated clinical studies which were registered with clinicaltrials.gov. It was approved by the ethics committee of the University of Göttingen, Germany.

Study population

The basic population consisted of all recruiting or planned investigator-initiated clinical studies registered with the study register of the U.S. National Institutes of Health (clinicaltrials.gov) in May 2006 (n = 7988). Using computer generated random numbers we selected 1500 studies. Investigator-initiated studies were defined as all studies which were registered with a non-industry sponsor. In July 2006, all principal investigators with valid contact details were contacted by e-mail explaining the purpose of the survey and a link to access the questionnaire. Two reminder e-mails were sent each two weeks apart. No attempts were made to contact non-responders by phone because of difficulties to obtain sufficient contact details and resource constraints. Written informed consent was not sought in this internet based survey but informed consent of participants was implied because access to the questionnaire was restricted to persons approached by e-mail explaining the study.

Data collection

A web-based 52-item questionnaire was developed based on the WHO Trial Registration Data Set [2] and the proposed minimum data set from the Ottawa Group [6]. Participants were asked whether they would be willing to publicly disclose specific details about their study. The questionnaire consisted of eleven main sections (Additional file 1): Contacts and funding, title and description of the study, key dates, ethical approval, background of the study, eligibility criteria, intervention(s), outcome measures, design, documents, and results. We used five response categories to indicate willingness for disclosure: yes, no, don't know/can't decide, don't want to answer, not applicable. Each question had to be answered in order to proceed with the questionnaire. Given that the questionnaire solely consisted of questions directly related to items of the two data sets mentioned above, no piloting of the questionnaire was done.

Sample size calculation

Sample size calculation was based on confidence intervals for proportions and the following considerations: 1) a maximum width of a confidence interval for questions answerable dichotomously of 6% was considered narrow enough; 2) based on a basic population of 7988 this would require 259 responses; 3) assuming a response rate of 25% a sample size of 1036 would be needed; 4) to account for missing and invalid contact details we selected 1500 studies.

Statistical analysis

Data were analysed in Stata 9.2 using descriptive statistics for the main analysis. Responders and non-responders were compared using the chi-squared test. Association between willingness to disclose study details and study phase was assessed using the chi-squared test for trend. To explore the potential influence of non-response bias we used logistic regression to assess associations between factors associated with non-response and the willingness to register study details. We therefore included the following factors in the regression analysis: study sponsored by the National Institutes of Health and phase II study in oncology. For ease of interpretation we recoded "yes" and "not applicable" responses as 'yes' and "no", "don't want to answer", and "don't know" as 'no'. We chose this coding procedure in order to predict willingness to disclose all study details according the Ottawa criteria. Coding "not applicable" as "no" would have had the following potential consequence: e.g. if a participant is willing to disclose the whole data-set except one item which is not applicable to his context, he would have been falsely categorized as completely reluctant (i.e. "0" in regression analysis).

Results

201 out of the initial 1500 studies had no (n = 184) or no valid (n = 17) e-mail address of the principal investigator available, leaving 1299 potential participants. Studies with principal investigators with no or invalid e-mail address were more likely to be sponsored by the National Institutes of Health, to be in the field of oncology, and more likely to study a drug. Of the 1299 contacted investigators, 282 (22%) responded and these formed the population for analysis (Figure 1).
Non-responders were more likely to be principal investigators of studies sponsored by the National Institutes of Health (Table 2). Studies of non-responders were also more likely to be in the field of oncology and registered as phase II studies. Given these differences in responders and non-responders, we did not calculate confidence intervals of response distributions.
Table 2
Study characteristics
Variables
Total
E-mail available*
E-mail not available
p**
Responders
Non-responders
p
Recruiting status of study
1500
1299
201
0.15
282
1017
0.68
   Recruiting
1375 (91.7)
1196 (92.1)
179 (89.0)
 
258 (91.5)
938 (92.23)
 
   Not yet
125 (8.3)
103 (7.9)
125 (8.3)
 
24 (8.5)
79 (7.8)
 
Type of sponsor
   
<0.001
  
<0.001
   NIH
288 (19.2)
207 (15.9)
81 (40.3)
 
21 (7.4)
186 (18.3)
 
   Other
1212 (80.8)
1092 (84.1)
120 (59.7)
 
261 (92.5)
831 (81.7)
 
Study condition according to ICD 10
   
<0.001
  
0.03
   Infections
69 (4.6)
64 (4.9)
5 (2.5)
 
22 (7.8)
42 (4.1)
 
   Oncology
608 (40.5)
474 (36.5)
134 (66.7)
 
73 (25.9)
401 (39.4)
 
   Haematology
23 (1.5)
19 (1.5)
4 (2.0)
 
3 (1.1)
16 (1.6)
 
   Endocrinologic diseases
85 (5.7)
79 (6.1)
6 (3.0)
 
21 (7.4)
58 (5.7)
 
   Psychiatric diseases
119 (7.9)
109 (8.4)
10 (4.9)
 
28 (9.9)
81 (8.0)
 
   Neurologic diseases
62 (4.1)
58 (4.6)
4 (2.0)
 
14 (5.0)
44 (4.3)
 
   Eye
12 (0.8)
2 (1.0)
10 (0.8)
 
2 (0.7)
8 (0.8)
 
   Ear
2 (0.1)
2 (0.1)
0 (0.0)
 
1 (0.3)
1 (0.1)
 
   Cardiovascular diseases
123 (8.2)
118 (9.1)
5 (2.5)
 
30 (10.6)
88 (8.6)
 
   Lung diseases
59 (3.9)
53 (4.1)
6 (3.0)
 
10 (3.5)
43 (4.2)
 
   Digestive diseases
43 (3.3)
4 (2.0)
47 (3.1)
 
15 (5.2)
28 (2.7)
 
   Skin
19 (1.3)
18 (1.4)
1 (0.5)
 
3 (1.1)
15 (1.5)
 
   Musculoskeletal disorders
51 (3.4)
48 (3.7)
3 (1.5)
 
12 (4.3)
36 (3.5)
 
   Urology
76 (5.1)
72 (5.5)
4 (2.0)
 
20 (7.1)
52 (5.1)
 
   Pregnancy
26 (1.7)
23 (1.8)
3 (1.5)
 
5 (1.8)
18 (1.8)
 
   Perinatal disorders
10 (0.7)
10 (0.8)
0 (0.0)
 
3 (1.1)
7 (0.7)
 
   Labour
2 (0.1)
2 (0.1)
0 (0.0)
 
1 (0.3)
1 (0.1)
 
   Trauma
39 (2.6)
35 (2.7)
4 (2.0)
 
6 (2.1)
29 (2.8)
 
   Extern
1 (0.1)
1 (0.1)
0 (0.0)
 
1 (0.3)
0 (0.0)
 
   Other
67 (4.5)
61 (4.7)
6 (3.0)
 
12 (4.3)
49 (4.8)
 
Study intervention
   
<0.001
  
0.10
   Drug
895 (59.7)
741 (57.0)
154 (76.6)
 
143 (50.7)
598 (58.8)
 
   Procedure
231 (15.4)
211 (16.2)
20 (9.9)
 
55 (19.5)
156 (15.3)
 
   Behaviour
112 (7.5)
107 (8.2)
5 (2.5)
 
32 (11.3)
75 (7.4)
 
   Device
64 (4.3)
62 (4.8)
2 (1.0)
 
14 (5.0)
48 (4.7)
 
   Vaccine
17 (1.1)
12 (0.9)
5 (2.5)
 
3 (1.1)
9 (0.9)
 
   Gene therapy
5 (0.3)
4 (0.3)
1 (0.5)
 
0 (0.0)
4 (0.4)
 
   Other/not applicable
176 (11.7)
162 (12.5)
14 (7.0)
 
35 (12.4)
127 (12.5)
 
Study phase
   
<0.001
  
<0.01
   Phase I
137 (9.1)
113 (8.7)
24 (11.9)
 
19 (6.7)
94 (9.2)
 
   Phase II
357 (23.8)
281 (21.6)
76 (37.8)
 
45 (16.0)
236 (23.2)
 
   Phase III
230 (15.3)
206 (15.9)
24 (11.9)
 
51 (8.1)
155 (15.2)
 
   Phase IV
177 (11.8)
168 (12.9)
9 (4.5)
 
48 (17.0)
120 (11.8)
 
   Phase I/II
72 (4.8)
56 (4.3)
16 (8.0)
 
6 (2.1)
50 (4.9)
 
   Phase II/III
43 (2.9)
35 (2.7)
8 (4.0)
 
6 (2.1)
29 (2.8)
 
   Not applicable
484 (32.3)
440 (33.9)
44 (21.9)
 
107 (37.9)
333 (32.7)
 
* E-mail address of principal investigator
** Chi-squared test
Categories with less than six counts overall were added to category "Other" or "Other/not applicable" for the chi-square test.
Agreement to disclose study details was very high for most individual items (Additional file 1). Willingness to register study details was lowest for details of planned subgroup analyses (49% willingness to register), sample size calculation (57%), and planned analyses methods (58%). However, disclosing study documents, in particular the study protocol and financial agreements, was found to be most problematic with only 31% of respondents willing to disclose these publicly. 151 investigators (54%) would be willing to present results of their studies in a register. However, only 89 participants (32%) would be willing to present results before submission to a peer-reviewed journal even if journal editors would accept such a presentation (a situation that would be comparable to today's practice regarding presentation at conferences).
Sixty-eight responders (24%) were willing to disclose all items required by the WHO – and clinicaltrials.gov. Only 34/282 (12%) agreed to disclose all details proposed by the Ottawa Group (Figure 2).
If disclosing documents would not be required, 67/282 (24%) participants would be willing to disclose all other details. Not requiring registration of planned analyses methods including subgroup analyses only slightly increased this proportion (70/282; 25%). Restricting the data set to study details not identified as sensitive for competitive reasons by industry (scientific title, intervention, sample size calculation, outcome measures) also increased this proportion only slightly (71/282; 25%). There was no association between study phase and willingness to register details according to the data set proposed by the Ottawa Group (test for trend p = 0.59). Logistic regression indicated no association between characteristics of non-responders (study sponsored by the National Institutes of Health and phase II study in oncology) and willingness to disclose details (p > 0.15 for each explanatory variable).

Discussion

This is the first survey on opinions of academic researchers suggesting that principal investigators of non-industry sponsored studies are reluctant to disclose all data items proposed by the Ottawa Group. Disclosing the study protocol and financial agreements were found to be most problematic.
As with any survey, our study is susceptible to non-response bias. The response rate was low with only 22% of potential participants responding. It is well known that web-based surveys have lower response rates than postal surveys [8]. Exact reasons for this low response rate remain unclear for us but might be related to the length of the questionnaire, the absence of any incentive, or the topic itself [9]. Nevertheless, we found no association between characteristics associated with non-response and the willingness to disclose study details. Although the impact of non-response seems therefore less problematic this study should be viewed as exploratory. Socially desirable responses might be another problem often encountered in surveys. Drawing conclusions about the actual habit of respondents is therefore often problematic [10]. Given the recent debate about trial registration and the requirement of editors of major journals to register a clinical study, one might expect that respondents would favour registration simply to conform with opinion leaders. The high proportion of respondents willing to disclose individual study details might be viewed as an indication for socially desired response habit. However, willingness to disclose all proposed items was low indicating that respondents differentiated well between different items. Finally, because of limited information available, we were not able to explore characteristics of investigators willing to disclose study details such as country of origin or host institution.
Recent debates about the WHO Trial Registration Data Set have focused on the reluctance of industry to disclose particular study details [11, 12]. Representatives of industry argue that some study details are sensitive for competitive reasons and that these details might be disclosed with some delay [13]. However, academic research is also competitive nowadays and researchers might therefore also be reluctant to disclose study details. This reluctance might be even more pronounced for study registers adopting the data set proposed by the Ottawa Group [6]. The main reasons for prospective trial registration correspond to publication and outcome reporting bias [14, 15]. For example, trial participants are at risk of being misled when their consent and the trial design are not fully informed by prior research or institutional review board (IRB) members cannot fully weigh risks and benefits when some unknown proportion of the relevant data is unavailable for review [16]. Consequently, it is an ethical obligation of researchers to publish all relevant details of their clinical studies and publicly available clinical study registers help to fulfil this obligation [17].

Conclusion

As our survey suggests, academic researchers might be reluctant to disclose details of their study. Future discussions on trial registration should not only focus on industry but also on academic researchers. Consequences of more detailed registration of studies should be openly discussed when considering a minimum data set for a given trial register. These potential consequences not only include issues related to commercial and academic competition but also the increasing administrative workload of clinical researchers. Lastly, it seems important to disseminate the ethical obligation to register clinical studies more broadly.

Acknowledgements

MS is supported by a Young Investigators' Award of the German Ministry of Education and Research (grant no. 01 GK 0516). We gratefully acknowledge Irene Kühne for her technical assistance.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://​creativecommons.​org/​licenses/​by/​2.​0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Competing interests

ST was a member of the Ottawa Group. MS declares that he has no competing interests.

Authors' contributions

MS and ST equally participated in study design, statistical analysis and writing of the manuscript. Both authors had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Both authors read and approved the final manuscript.
Zusatzmaterial
Additional file 1: All questions and corresponding results. The data provide represent the results of the survey. (DOC 136 KB)
12913_2007_586_MOESM1_ESM.doc
Authors’ original file for figure 1
12913_2007_586_MOESM2_ESM.pdf
Authors’ original file for figure 2
12913_2007_586_MOESM3_ESM.pdf
Literatur
Über diesen Artikel

Weitere Artikel der Ausgabe 1/2008

BMC Health Services Research 1/2008 Zur Ausgabe