Individual participant data meta-analysis tips and tricks: troubleshooting commonly encountered issues of contacting trialists for individual participant data

Abstract

Objective:

We present a framework to guide researchers in contacting and retrieving trial data from trialists. This framework serves 2 purposes: i) to provide a consistent and transparent approach for contacting authors, and ii) to describe how to record clearly all contact attempts and to identify trialists who have not responded to reasonable attempts at communication. This framework will help researchers identify trials that may require investigation for data integrity issues.

Background:

Individual participant data meta-analysis (IPD-MA) is considered the gold standard for evaluating clinical interventions. The popularity of IPD-MA has increased due to the potential for advanced statistical analyses and the ability to test data veracity prior to analysis. Contacting trialists and requesting data is the most time-intensive step in an IPD-MA project. Often, many datasets are not retrieved, as authors are uncontactable or do not share data. This absence of IPD can bias meta-analysis and interpretation of results. Currently, there is no framework in place to guide researchers in contacting trialists and to define a reasonable point to cease communication attempts.

Proposed framework:

The framework consists of 4 approaches: first, contacting the listed authors on the trial publication; second, contacting the trialists’ associated institutions (hospitals and universities); third, contacting colleagues from similar regions within a particular country; and fourth, contacting the journal in search of trialists’ contact details. If trialists do not respond to sustained communication attempts, their study should be classified as “non-responding.” Depending on the trial context, non-responding trialists or those who respond with concerning reasons why data are unavailable may be subject to further review regarding trial quality and data integrity concerns.

Case study:

This framework was applied in an IPD-MA comparing misoprostol with oxytocin for the prevention of postpartum hemorrhage, which included 79 randomized controlled trials. With the use of this framework, trialists from 10 trials responded to the IPD invitation and contributed data (6 of which were used in final analysis); 38 trialists responded but did not contribute data; and 31 trialists did not respond and their trials were subsequently classified as non-responding.

Conclusions:

The proposed framework provides a uniform structure for contacting authors and requesting data for IPD-MA, which may increase the likelihood that trialists will respond to IPD-MA invitations. This framework will also help to identify trialists who do not respond to reasonable attempts at communication.

Introduction

Evidence-based medicine (EBM) centers on the judicious use of current best evidence to inform clinical decisions for individual patients.1 Randomized controlled trials (RCTs) are regarded as the gold standard for comparing medical interventions and are integral to EBM.2 However, for many interventions it is labor-intensive, time-consuming, and expensive to collect sufficient data to demonstrate a clinically important effect. Meta-analysis (MA) offers an alternative method for acquiring and analyzing the large amount of data that would, ideally, be generated by a single, well-resourced RCT. MA performed on original trial data is known as individual participant data meta-analysis (IPD-MA).

There are many well-established benefits of IPD-MA. Broadly, IPD-MA reduces research waste, increases data sharing, and allows for personalized health care.3,4 Specifically, access to IPD allows advanced analyses due to increased statistical power and the ability to standardize exposure and outcome definitions.5 Furthermore, access to IPD allows interrogation of data trustworthiness, which is necessary amid increasing evidence of data fabrication within medicine.6,7 IPD can be analyzed for unusual data patterns, baseline characteristics, patterns of allocation, and internal and external consistencies.8

Due to this litany of benefits, IPD-MA projects are increasing in popularity. For the most part, IPD methodology is described in detail.4,9 However, there is a paucity of literature specifically guiding researchers through the process of contacting, inviting, and subsequently requesting trialists’ IPD, which is arguably one of the most time-consuming steps in IPD-MA. The process of contacting and seeking IPD from trialists is important, as most IPD-MA projects include only a fraction of the available data, which may bias MA results and their interpretation. Historically, this step is poorly documented.10 A consistent approach may increase the number of trialists who respond and subsequently share data or, in the event of non–data sharing or non-response to IPD invitation, allow researchers to document transparently the reasons for not sharing data and noting which trialists did not respond.

Traditionally, the inability to contact trialists and obtain IPD has been attributed to foreseeable problems with data sharing or ethics agreements, or due to a historic trial where the data are reasonably unavailable.11,12 However, due to the increasing evidence of untrustworthy data within medicine,6,7 the absence of trialist response and/or data sharing deserves closer analysis, as it may indicate a trial with compromised data integrity.

To our knowledge, there are no existing protocols to guide researchers in contacting and retrieving data from trialists. Here, we present a framework that serves 2 purposes: i) to provide a consistent and transparent approach for contacting authors, and ii) to provide a method to clearly record all contact attempts, and to identify trialists who have not responded to reasonable attempts at communication. This framework will help researchers identify trials that may require investigation for data integrity issues.

We present the application of this framework to a case study, an IPD-MA comparing misoprostol with oxytocin for preventing postpartum hemorrhage.13

Methodological approach

Finding contact details for authors

Finding current contact details for authors can be difficult. Our proposed checklist for finding email addresses and telephone numbers for authors is detailed in Table 1. Publications or trial registrations typically designate 1 trialist as the corresponding author, which is usually accompanied by their email address or telephone number. This should be the first point of contact for an email or telephone invitation to contribute IPD. Other email addresses for the corresponding author or any additional trialists may appear in the trial registration (if the RCT is registered), Open Researcher and Contributor ID (ORCID) database, or other publications. We suggest individually searching each trialist in medical research databases (eg, PubMed or Ovid MEDLINE) and checking all publications for email addresses. Located email addresses may belong to the trialist of interest or a colleague with whom that trialist has previously published.

Table 1.

Example checklist and template for locating trialists’ contact details when conducting an individual participant data meta-analysis

The next point of contact should be a search for trialists’ profiles on social media platforms, including ResearchGate, LinkedIn, Facebook, and X (formerly Twitter). Occasionally, email addresses may be listed on trialists’ profiles; alternatively, they can be messaged on these platforms. If trialists have a private practice as clinicians, they may have a personal or business website. These may be found in online search engines using search terms such as trialist name + country + region + field of specialization (eg, Jane Smith + Australia + Melbourne + Obstetrics and Gynacology).

Finally, a trialist may consult for a private company or hold an advisory position on a committee, which will provide a further opportunity to identify alternative email addresses to which an invitation may be sent.

If these strategies do not yield useful contact information, the next stage is to email the relevant head of department at the hospitals and universities with which the trialist is affiliated. It may be possible to locate their contact details on official institutional websites. If email addresses are not listed, the process described previously may be applied to gain contact details for the head of department, departmental professors, or clinicians associated with the trialist in question. The same approach can be used to contact colleagues of trialists identified through joint publications or institutional websites. Contact details for funding bodies, such as pharmaceutical companies, can be found online at the company’s website.

Contacting authors with data requests

Researchers can begin sending invitations to participate in an IPD-MA once they have formulated a list of RCTs for inclusion. A flowchart of the order of author inquiry approaches is presented in Table 2. A first-round IPD-MA invitation should be sent to the corresponding author’s email listed in the publication (Approach 1a). If there is no response after 2 weeks, a second-round invitation email should be sent to the corresponding author and the first author (Approach 1b). Similarly, if there is no response after a further 2 weeks, a third-round email should be sent to the corresponding author, first author, and last author (Approach 1c). A fourth-round email to all listed authors can be sent if there is no response after another 2 weeks (Approach 1d).

Table 2:

Flowchart of author inquiry for randomized controlled trials for individual participant data and labeling the “non-shared” status

Approach 1a	• Contact the corresponding author.
	• Go to Approach 1b if the email bounces or no response after 2 weeks.
Approach 1b	• Email the corresponding author and first author (if these are different authors).
	• Go to Approach 1c if the email bounces or no response after 2 weeks.
Approach 1c	• Email the corresponding author, first author, and last author.
	• Go to Approach 1d if the email bounces or no response after 2 weeks.
Approach 1d	• Email all co-authors listed on the publication.
	• Go to Approach 2a if the email bounces or no response after 2 weeks.
Approach 2a	• Contact the affiliated institutions (hospitals and universities) and funding bodies.
	○ Email the head of department or secretary to the head of department; ask for contact details of all authors.
	○ Email departmental professors; ask for contact details of authors and contact details for the head of department.
	○ Email departmental administration; ask for contact details of authors and contact details for the head of department.
	○ If there is no department-specific webpage, email hospital/university administration; ask for contact details for department or the head of department.
	○ Email the funding bodies (eg, industry sponsors, pharmaceutical companies).
	• Concurrently continue to email all co-authors listed on the publication every 2–6 weeks.
	• Go to Approach 2b if there is no response after 2 weeks.
Approach 2b	• Email colleagues of authors
	○ who have published with authors recently (ie, on PubMed or ResearchGate)
	○ from same hospital/university
	○ from same country who you may be emailing regarding a different study.
	• Concurrently, continue to email all co-authors listed on the publication every 2–6 weeks.
	• Go to Approach 3 if there is no response after 2 weeks.
Approach 3	• Email network from the relevant state or country and ask for assistance in contacting the authors.
	• Go to Approach 4 if no response or they cannot help (email every 2–4 weeks).
	• Add your study to the relevant shared document (if there is a coordinated communication approach for the country of study).
Approach 4	• Contact the relevant head of department. If there is no email address for the head of department, the next step is to email departmental professors (more senior members of staff are likely to have been at the institution longer, increasing the probability that they may know the author of interest) or administration.
	• Consider contacting the journal editor as the last step; label the study as “non-responding” if no new information is revealed in this final step.

If there has been no response after Approaches 1a–1d (Table 2), researchers should proceed to Approach 2a: contacting institutional affiliations. Commonly, authors are affiliated with a university or a hospital. If it is not possible to email senior staff members, email the university and/or hospital administration asking for contact details of both the authors of interest and the head of department. Funding bodies, including industry sponsors such as pharmaceutical companies, should be emailed. Researchers can use their discretion to determine the appropriate number of times to re-email institutional contacts and funding bodies. If there is no response from any of the institutional contacts 2 weeks after the last email was sent, proceed to Approach 2b.

Approach 2b requires locating colleagues of the RCT author group. Colleagues may be identified through institutional websites (as previously mentioned) or joint publications. Furthermore, a researcher may be in contact with authors of a different study from a similar region or country as the study of interest; in this situation, one may email the trialists of the other study and ask if they know the trialists of interest.

In Approach 3, the researcher may ask colleagues to liaise with relevant international contacts from their personal network. For example, a researcher may have colleagues with international affiliations who have institutional connections in the same country or region as the trialists in question. These contacts may need to be prompted to respond, with a recommended time frame for prompting contacts every 2-4 weeks; if no response is obtained, researchers may progress to Approach 4. If possible, add the RCT to a list of trials organized by country. This may subsequently provide an opportunity to contact an academic from a particular country and ask if they recognize any trialists from this compiled list. These contacts may need to be prompted to respond, with a recommended time frame for prompting contacts every 2-4 weeks; if no response is obtained, researchers may progress to.

Labeling an RCT as “non-responding”

If there is no reply after prolonged, repeated attempts to contact the trialists and their colleagues from associated universities and hospitals, researchers may progress to Approach 4 (Table 2). As a final step, they may contact the journal or publication and request the authors’ contact details. This may yield new contact information and prompt further attempts to contact authors by email. If the journal does not respond or no new contact information is provided, the RCT is then classified as “non-responding.”

A trialist response to IPD-MA invitation should be clearly documented. The trialist may agree to participate in the IPD but may subsequently stop responding to emails with requests for IPD. If trialists continue to not respond to emails after this point, researchers may label this trial as “non-responding.” Alternatively, if trialists respond stating that IPD cannot be shared, researchers should document this and consider investigating the trialists’ response. For example, if trial data are unavailable due to them being lost or destroyed, the researcher may choose to enquire how or why this occurred, especially if the trial was conducted recently (researchers must exercise judgment when defining recent; eg, if a trial recruited participants within the past 5 years and the data are no longer available, this is concerning).

Case study findings

Recently, we conducted an IPD-MA comparing oxytocin with misoprostol for the prevention of postpartum hemorrhage.13 Ovid MEDLINE, Ovid Embase, Ovid Emcare, CINAHL Plus (EBSCOhost), Scopus, Cochrane Central Register of Controlled Trials, and ClinicalTrials.gov were searched for peer-reviewed literature, meeting the inclusion criteria of RCTs comparing misoprostol with oxytocin for the prevention of postpartum hemorrhage. After screening, 79 RCTs were identified, and trialists were contacted over 24 months, from April 2021 to April 2023. Trialists from 48 of 79 trials (60.8%) responded to the IPD-MA invitation (Table 3), while trialists from the remaining 31 trials (39.2%) did not respond. Of the trialists who responded, 10 (20.8%) shared IPD and 38 (79.2%) did not. Six of 10 IPD were able to be used in IPD-MA, but 4 IPD were excluded due to trustworthiness concerns.

Table 3:

Case study—trialists’ responses to invitations for individual participant data meta-analysis project: misoprostol vs oxytocin for the prevention of postpartum hemorrhage14

Overview of trialists’ responses to individual participant data meta-analysis invitation
Non-responders	N = 31/79 (39.2%)
Responders	N = 48/79 (60.8%)
Response = Yes (willing/able to participate)	N = 10 (12.7% of the total and 20.8% of responders)
Response = No (unwilling/unable to participate)	N = 38 (48.1% of the total and 79.2% of responders); see Table 4 for list of reasons supplied for declining invitation
Overview of trialists’ responses by year of trial publication (trials that responded/total trials)
2005 or before	16/18 (88.9%)
2006–2010	11/12 (91.7%)
2011–2015	12/23 (52.2%)
2016–2020	5/15 (33.3%)
2021 or after	4/11 (36.4%)

Of the 48 trials where a trialist responded, 41 designated a corresponding author and 7 did not. The corresponding author responded to the IPD-MA invitation in 23 of 41 (56.1%) instances. Of these 23 responses, only 7 (30.4%) corresponding authors were contactable via the details listed in the publication. The remaining 16 (69.6%) corresponding authors were contacted via other avenues, including other email addresses (10/23, 43.5%; provided by colleagues [n=4], located from other publications [n=3], or located on institutional websites [n=3]); WhatsApp invitations sent to mobile phone numbers (5/23, 21.7%; located on institutional websites [n=3] or other publications [n=2]); or through ResearchGate (1/23, 4.3%). In 18 of 41 (43.9%) instances, the IPD-MA invitation response was from an author on the publication who was not the designated corresponding author.

The response rate from trialists varied according to year of publication (Table 3). Regarding RCTs published in 2005 or earlier, the response rate was 88.9% (16/18). The response rate was similar (91.7%; 11/12) for RCTs published between 2006 and 2010. The response rate decreased for more recent studies. For RCTs published between 2011 and 2015, the response rate was 52.2% (12/23). For RCTs published between 2016 and 2020, the response rate was 33.3% (5/15). For RCTs published after 2020, the response rate was similar at 36.4% (4/11).

An overview of the 79 RCTs and their responses to IPD-MA invitation is summarized in Table 4.

Table 4:

Overview of all studies included in the case study: individual participant data meta-analysis comparing misoprostol vs oxytocin for the prevention of postpartum hemorrhage14

#	Author	Year	Country	# Patients	Author response	Outcome	Explanation
1	Acharya	2001	UK	60	Yes	Declined invitation	Data securely destroyed in 2021
2	Afkham	2022	Iran	128	Yes	Accepted invitation	IPD received (IPD excluded from MA)
3	Afolabi	2010	Nigeria	200	Yes	Declined invitation	Data lost
4	Al-Sawaf	2013	Egypt	67	No	No response
5	Alwani	2014	India	200	No	No response
6	Amin	2014	Pakistan	200	Yes	Declined invitation	Data lost
7	Andanikin	2012	Nigeria	218	Yes	Declined invitation	Data with PI, responding author does not have access
8	Andanikin	2013	Nigeria	50	Yes	Declined invitation	Data with PI, responding author does not have access
9	Anita	2018	India	126	No	No response
10	Asmat	2017	Pakistan	1678	Yes	Declined invitation	Data lost
11	Atukunda	2014	Uganda	1140	Yes	Accepted invitation	IPD received
12	Baskett	2007	Canada	622	Yes	Declined invitation	Data lost
13	Begum	2015	Bangladesh	100	Yes	Declined invitation	Data lost in hospital flooding
14	Bellad	2012	India	652	Yes	Declined invitation	Data lost
15	Benchimol	2001	France	600	Yes	Declined invitation	Data lost
16	Bhatti	2014	Pakistan	120	No	No response
17	Bugalho	2001	Mozambique	700	Yes	Declined invitation	Data lost (RCT >20 years old)
18	Burman	2021	India	80	Yes	Accepted invitation	IPD received
19	Caliskan	2002	Turkey	803	Yes	Declined invitation	Data lost (RCT >20 years old)
20	Caliskan	2003	Turkey	772	Yes	Declined invitation	Data lost (RCT >20 years old)
21	Chaudhuri	2010	India	190	Yes	Declined invitation	Ethics expired, not permitted to share data
22	Chaudhuri	2012	India	530	Yes	Declined invitation	Ethics expired, not permitted to share data
23	Cook	1999	Australia/ PNG/China	863	No	No response
24	Dabbaghi Gale	2012	Iran	269	Yes	Declined invitation	Data lost
25	Dasuki	2002	Indonesia	196	No	No response
26	Diallo	2017	Senegal	304	No	No response
27	Dutta	2016	India	400	No	No response
28	Eftekhari	2009	Iran	100	Yes	Declined invitation	Data not kept
29	Elbohoty	2016	Egypt	175	No	No response
30	El-Refaey	2000	UK	1000	Yes	Declined invitation	Data lost (RCT >20 years old)
31	Fakour	2013	Iran	200	No	No response
32	Fazel	2013	Iran	100	No	No response
33	Gavilanes	2015	Ecuador	200	Yes	Accepted invitation	IPD received
34	Gerstenfeld	2001	US	325	Yes	Declined invitation	Data lost (RCT >20 years old)
35	Ghafoor	2022	Pakistan	100	Yes	Declined invitation	Authors do not want to share data
36	Gulmezoglu	2001	Many (WHO)	18530	Yes	Declined invitation	Data unavailable from WHO
37	Gupta	2006	India	200	Yes	Accepted invitation	IPD received
38	Jain	2019	India	48	No	No response
39	Karkanis	2002	Canada	223	Yes	Declined invitation	Data lost (RCT >20 years old)
40	Kaudel	2020	Nepal	120	Yes	Declined invitation	Data lost
41	Kundodyiwa	2001	Zimbabwe	499	Yes	Declined invitation	Data lost
42	Lokugamage	2001	UK	40	Yes	Declined invitation	Data securely destroyed in 2021
43	Lumbiganon	1999	South Africa/ Thailand (WHO)	597	Yes	Declined invitation	Data unavailable from WHO
44	Mishra	2023	India	407	No	No response
45	Mitwaly	2022	Egypt	77	No	No response
46	Modi	2014	India	50	No	No response
47	Mukta	2013	India	200	Yes	Accepted invitation	IPD received (IPD excluded from MA)
48	Musa	2015	Nigeria	200	Yes	Accepted invitation	IPD received
49	Najeeb	2022	Pakistan	306	No	No response
50	Nankaly	2016	Iran	185	No	No response
51	Narrey	2023	India	100	No	No response
52	Nasr	2009	Egypt	514	Yes	Declined invitation	Too busy to participate
53	Ng	2004	Hong Kong	309	Yes	Accepted invitation	IPD received
54	Oboro	2003	Nigeria	496	Yes	Not participating	Data not kept
55	Othman	2016	Egypt	120	No	No response
56	Owonikoko	2011	Nigeria	100	No	No response
57	Pakniat	2015	Iran	100	No	No response
58	Parsons	2006	Canada	450	Yes	Not participating	Ethics expired
59	Parsons	2007	Canada	440	Yes	Not participating	Ethics expired
60	Patil	2023	India	200	No	No response
61	Penaranda	2002	Colombia	50	Yes	Declined invitation	Data lost (RCT >20 years old)
62	Perez Rumbos	2017	Venezuela	392	Yes	Accepted invitation	IPD received (IPD excluded from MA)
63	Rajaei	2014	Iran	400	No	No response
64	Rawat	2021	India	240	No	No response
65	Roy	2017	India	200	No	No response
66	Sadiq	2011	Nigeria	1800	Yes	Declined invitation	Data unavailable
67	Satyajit	2017	India	100	No	No response
68	Shady	2017	Egypt	240	No	No response
69	Shah	2021	Pakistan	334	Yes	Accepted invitation	IPD received (IPD excluded from MA)
70	Shaheen	2019	Pakistan	212	No	No response
71	Shrestha	2011	Nepal	200	No	No response
72	Singh	2009	India	225	Yes	Not participating	Data unavailable
73	Snehalata	2023	India	120	No	No response
74	Sultana	2007	Bangladesh	310	No	No response
75	Tewatia	2014	India	100	Yes	Declined invitation	Data lost
76	Vagge	2014	India	200	Yes	Declined invitation	Not interested
77	Vimala	2006	India	100	Yes	Declined invitation	Data lost
78	Walley	2000	Canada	392	Yes	Declined invitation	Ethics expired
79	Zachariah	2006	India	1347	Yes	Declined invitation	Data lost

IPD, individual participant data; MA, meta-analysis; PI, principal investigator; RCT, randomized controlled trial; WHO, World Health Organization

Discussion

Strengths and limitations

Our proposed flowchart (Table 2) guides researchers through the fraught process of contacting trialists and asking them for data. This framework was developed within the Department of Obstetrics and Gynaecology at Monash University in Melbourne, Australia, through discussion with both senior and junior clinicians and researchers, and through collaboration with colleagues working at other Australian and international universities. Many of these collaborators were conducting IPD-MA projects and had trouble locating contact details for trialists and subsequently retrieving IPD. This framework has been applied to several IPD-MA projects from our research department.13-16 This flowchart outlines the different avenues that researchers may take to find contact details for trialists and defines a reasonable limit for this process. The acceptable limits of this flowchart were determined through discussion with academic researchers who have many years of experience with contacting trialists for data.

There are many considerations for applying this framework. First, this framework is a guide and not a one-size-fits-all approach. Researchers will need to apply discretion when considering how many times to email trialists and the time between emails. Similarly, the trial context is important. If researchers have received no response from trialists regarding a historic trial, subsequent approaches in the framework may be omitted; for example, they may choose not to pursue the affiliated institutions or journal of publication.

Second, the framework favors email as the primary form of communication. If countries or cultures communicate differently, this framework is less easily applied. For example, we found that many South Asian trialists were less likely to respond over email and more likely to respond over the telephone. This may render subsequent approaches in the framework less relevant, as institutions and journals may be hesitant to disseminate trialists’ phone numbers. Furthermore, we attempted to contact authors in English. If authors do not speak English, then they may not respond to invitations.

Importantly, this framework aims to elicit a response from trialists; it does not assess the authors’ response. If this framework is implemented in the practice of collating and analyzing IPD, it may increase the number of authors who respond but may not increase the number from whom IPD is retrieved. Authors may respond with predictable excuses for not sharing IPD, and the original problem of low rates of data sharing remains.

Comment

The framework we propose for contacting trialists and requesting IPD serves 2 purposes: i) to provide a consistent and transparent approach for contacting trialists, and ii) to provide a stepwise approach to clearly identify and document trialists who have not responded to reasonable attempts at communication.

In our IPD-MA experience, this framework is necessary, as many corresponding authors are generally uncontactable via contact details listed on the trial publication, and trial data were sought from other members of the trial team. Without a clearly defined framework, it is difficult to decide when to persist and when to cease communication attempts. This framework provides a uniform approach. On several occasions, it seemed unlikely the trialists would respond; however, once an in-use email address or an active phone number was located, IPD was retrieved. Maximizing IPD retrieval from relevant trials is important for IPD-MA rigor and to avoid bias in results, which can compromise interpretation.5,17

A preexisting flowchart for guiding researchers in acquiring IPD was proposed by Veroniki et al.18 This flowchart highlights common communication difficulties with trialists and suggests contacting the authors 5 times over a period of 4 months. Veroniki et al.18 suggest emailing the corresponding author and, if the email is not working, to proceed with contacting other listed authors. In the event of no response, researchers should proceed to calling the corresponding author and, finally, to contacting the industry sponsor. While this framework provides a good overview, it lacks necessary detail, assuming that the corresponding author and the published email address are the best way to contact trialists. In theory, this should be true; however, in our experience, few invitation responses are received from the corresponding author via the published email address. Often, an email address appears to be working but the inbox is likely unmonitored; therefore, further guidance on other communication avenues is necessary. Additionally, the flowchart from Veroniki et al.18 does not adequately describe subsequent steps if the authors or industry sponsors are non-responding. Our flowchart, which builds on their approach, provides further practical advice on communication avenues as well as a defined end point to reasonable communication attempts.

In our case study, despite years of attempted communication, more than one-third of trialists were uncontactable. This is an important finding, as trialists whose work has been published recently should be contactable regarding published trial results. Logistical explanations provide an obvious excuse for no response, as trialists move institutions or leave academia. However, trialists of RCTs that were 20 years old were often easier to contact than trialists of RCTs published after 2020 (Table 3), which is an alarming finding. This may be explained, in part, by the geographical dissonance between these 2 groups of trials. Of 18 RCTs published prior to 2002, the most represented countries/organizational bodies were the United Kingdom, the World Health Organization, Turkey, and Canada (10/18, 55.6%). Of 11 RCTs published from 2021 onward, 100% came from India, Pakistan, Egypt, or Iran.

The difference in geographic distribution of these RCTs may be associated with a difference in communication; perhaps the latter countries do not routinely use email. Nonetheless, a lack of response from recently published trials (eg, published within 5 to 10 years) remains a concerning finding. There is an established connection between data availability and the trustworthiness of data19; we cannot discount that trialists may be non-responding because they are unwilling or cannot share their data. These trials should be screened for data integrity concerns, and there are many recently developed tools for this purpose.20-22

Due to a recent increase in IPD-MA popularity, there is a vast amount of literature on IPD-MA conduct. These guides often elucidate the laborious nature of contacting and communicating with trialists but do not troubleshoot this process in detail.4,23 Furthermore, reporting of this process in IPD-MA publications is notoriously poor; a 2015 IPD methodology review found that most IPD-MA projects do not report how data were acquired, why data were unavailable, or from how many trials and participants data were sought.10 This process should be transparent, especially considering that a majority of IPD-MA projects include only a fraction of supposedly available data,18,24 exposing the study to data availability bias, which can compromise the interpretation of results.25

Difficulty contacting trialists is not a new phenomenon. Similarly, retrieving IPD is a historically fraught process.23 To combat low rates of data sharing, many journals instigated mandatory data sharing statements declaring that data are “available upon request.” In 2023, Hussey11 examined the success of this policy and concluded, firstly, that “it is often not possible to correspond with corresponding authors”^(p.5) and, secondly, that trialists willingly include false statements, declaring that data are available upon request when, in fact. they are not.

Rates of data sharing have been consistently low. In 1962, a study found that 24% of data were available on request; in 2006, another study estimated that number to be 24.5%; and, in 2023, a systematic review estimated the range of data sharing was 0% to 37%.26-28 These low rates have been previously explained by logistical factors and institutional protocols preventing data sharing.23 However, given the growing evidence that there is a large amount untrustworthy data within medicine,29 trialist non-response and a lack of data sharing deserve closer examination. After all, there are plausible motivations to avoid sharing IPD. Original trial results are rarely reproducible, and data fabrication is much easier to detect with IPD.6,30 It is important to question whether some trialists refuse to share data due to fabrication or indefensible results.

Conclusion

The proposed framework is not a panacea for contacting authors and retrieving IPD. Many trialists will remain uncontactable, and many datasets will not be shared. However, this framework guides researchers to reasonably exhaust the avenues to contact trialists. This may increase IPD retrieval or result in a clear response as to why data are unavailable. Alternatively, it will transparently identify which trialists are uncontactable after reasonable communication attempts. This may prompt researchers to apply further measures to evaluate RCTs for data integrity concerns.

Funding

MF is supported by a Research Training Stipend, provided by the Australian Government.

BWM declared grants from NHMRC, personal fees from ObsEva, personal fees from Merck, personal fees from Guerbet, other from Guerbet, and grants from Merck, outside the submitted work.

DLR declared a grant from NHMRC and has received travel support and lecture fees from the International Society of Ultrasound in Obstetrics and Gynecology (ISUOG), unrelated to this work.

WL declared a grant from NHMRC that supports this work and received research grant funds from the Norman Beischer Medical Research Foundation, which was unrelated to this work.

None of the funding sources had any role in the design, development, or results of this study.

Author contributions

MF principally managed the project and the collaborative process, and carried out design, writing, editing, finalizing, and submitting of the manuscript. BWM conceived the research idea. All authors were involved in editing and finalizing the manuscript, and in the decision to submit the manuscript.