How transparent are GlaxoSmithKline’s clinical trials?

GlaxoSmithKline

Registration 296/500
(59.2%)
Summary results 270/500
(54%)
Clinical study reports 205/500
(41%)
Individual patient data 270/500
(54%)

Transparency rank

1st

Est. sales (USD)

$35bn

Known for

Seretide/Advair, Triumeq, Pediarix Vaccine

The results of clinical trials are routinely and legally withheld from doctors, researchers, and patients. We have assessed every company’s policy on clinical trials transparency to see whether they meet a series of standard transparency criteria, and checked our findings with the companies. We have then applied a score to each commitment, and used these to rank the companies. You can see all our raw data, and review our methods, on this site.

Below, you can read in more detail about the specific commitments made by this company on each of the four domains: registration, summary results, CSRs, and IPD.

8 out of 46

Registration

Why Registration is important

Before a trial is started, it should be “registered”: a description of the trial is posted on a publicly accessible trial registry, recognised by the World Health Organisation. This means that everyone knows the trial is happening, and will know if it is left unreported. It also helps readers detect if the trial has been designed, analysed, and reported correctly. This graph shows how GlaxoSmithKline’s policy on registration compares to others’.

Measures

We assess every company’s policy by reading through their commitments, and checking off whether they meet a range of criteria. We send every company our assessment, then apply a scoring system to each element. You can read about the scoring system here. You can read our detailed assessment of whether GlaxoSmithKline meets each of our criteria, below.

registration_1: Do they have a policy to register all trials (excepting specific exclusions in later columns) from now?
Yes / but only commit to on their own site, open q of whether this constitutes registration
registration_2: Do they say they conduct any kind of audit of compliance with their registration policy? (If they don't mention it, then no.)
No
registration_3: If they do conduct an audit of compliance, do they share the summary results of this audit publicly?
No
registration_4: If they do conduct an audit of compliance, do they share the line by line individual trial data from this audit publicly? (That is: the names of trials, and then whether they were in compliance with the policy, or not.)
No
registration_5: Does the policy include phase 4 trials? If they say "all" then this is assumed to include phase 4.
Yes
registration_6: Does their current policy describe the registration policy covering past trials?
Yes
registration_7: From what date does this policy apply? (Give explanation of how year is used eg "trials initiated after 2001" or "trials used in regulatory approval since 2003" etc, or if not other details given then "200x NOS" [not otherwise specified]).This date is normalised after the second "/" to permit comparisons between companies. The average duration of a clinical trial is 2 years, as per Pregelj 2015. Therefore we normalise to NOS date. If a company commitment is "trials initiated after 2010" then this becomes "2011", if the commitment is "trials completing after 2012" then this becomes "2011", and so on. Whole years are used for simplicity.
2004 nos / 2004

5 out of 46

Summary results

Why summary results are important

It is important that doctors, researchers, and patients have access to the summary results of every trial. These are the kinds of results you would typically find in an academic journal article: how the people in each arm of the trial did, over the course of the trial, measured by various criteria. This graph shows how GlaxoSmithKline’s policy on sharing summary results compares to others’.

Measures

We assess every company’s policy by reading through their commitments, and checking off whether they meet a range of criteria. We send every company our assessment, then apply a scoring system to each element. You can read about the scoring system here. You can read our detailed assessment of whether GlaxoSmithKline meets each of our criteria, below.

results_1: Do they have a policy to make all summary results available? (Excepting specific exclusions in later columns).
Yes
results_2: Do they commit to post summary results on pre-specified primary and secondary outcomes to clinicaltrials.gov within 12 months of completion? (Here, we are strict on 12 months, because without a time commitment there is effectively no commitment; where there is a longer time commitment, we give no and specify the delayed time period in a comment. If a company only says it commits to comply with legislation they get "no" here, as legislation does not cover all trials.)
No
results_3: Do they commit to post summary results to their own website within 12 months of completion? (Note same conditions on "within 12 months" apply as for posting results on clinicaltrials.gov).
No
results_4: Do they commit to submit all trial results to an academic journal within 12 months of completion. (This cannot be with caveats, eg "we submit all medically important results" scores "no").
No
results_5: Does this commitment to post summary results include unlicensed treatments?
Yes / "results from completed clinical studies on compounds that are both investigational and approved medicines"
results_6: Does this commitment to post summary results include unlicensed uses of licensed treatments? (In the absence of a clear commitment either way: if the company does not post summary results on unlicensed treatments, the answer on unlicensed uses of licensed treatments is assumed to also be "no"; if the company has a clear theme of "all" trials throughout their policy, then the answer is assumed to be "yes").
Yes
results_7: Does this commitment to post summary results include phase 4 trials? (In the absence of a clear commitment either way, if there is a clear theme of "all" throughout the policy document, for example if they have made commitments to phase 2-4 trials for other aspects of their policy and there is no reason to believe that this issue would be an exception to that, then this is "yes").
Yes
results_8: Does their current policy cover results of past trials, committing to make all results available (excepting specific exclusions in later columns)? If this commitment is only for a poorly defined subset of trials, such as "medically important results", this is coded as "no".
Yes
results_9: Do they commit to post summary results of all past trials (excepting specific exclusions in later columns) on pre-specified primary and secondary outcomes to clinicaltrials.gov? (Note this does not require a 12 month criteria for posting results).
No
results_10: Do they commit to post summary results of all past trials (excepting specific exclusions in later columns) to their own website? (Again, the retrospective commitment does not include a "within 12 months" requirement).
Yes
results_11: Do they commit to submit all trials to an academic journal. (Note this must be all trials, not "all interesting trials" etc).
Yes
results_12: Does this commitment to posting summary results of past trials include trials on unlicensed treatments?
Yes
results_13: Does this commitment to posting summary results of past trials results include unlicensed uses of licensed treatments? (In the absence of a clear commitment either way: if the company does not post summary results on unlicensed treatments, the answer on unlicensed uses of licensed treatments is assumed to also be "no"; if the company has a clear theme of "all" trials throughout their policy, then the answer is assumed to be "yes").
Yes
results_14: Does this commitment to post summary results of past trials include phase 4 trials? (In the absence of a clear commitment either way, if there is a clear theme of "all" throughout the policy document, for example if they have made commitments to phase 2-4 trials for other aspects of their policy and there is no reason to believe that this issue would be an exception to that, then this is "yes").
Yes
results_15: From what date does this policy on posting summary results for all trials apply? (Give explanation of how year is used eg "trials initiated after 2001" or "trials used in regulatory approval since 2003" etc, or if not other details given then "200x NOS" [not otherwise specified]).This date is normalised after the second "/" to permit comparisons between companies. The average duration of a clinical trial is 2 years, as per Pregelj 2015. Therefore we normalise to NOS date. If a company commitment is "trials initiated after 2010" then this becomes "2011", if the commitment is "trials completing after 2012" then this becomes "2011", and so on. Whole years are used for simplicity. For "drugs approved after year x" an additional year is added. So "approved after 2013" would normalise to "2011". There is no perfect method to make dates comparable between companies.
2004 nos / 2004

4 out of 46

Clinical study reports

Why Clinical Study Reports are important

Clinical Study Reports (CSR) are large detailed documents, sometimes thousands of pages long, which contain a wealth of detail on the methods and results of a trial. That information is often missing from other sources: one recent study estimates that CSRs contain twice as much information on benefits and harms as academic papers on trials. CSRs are routinely created for industry trials, but are less well known in the academic community. They follow a standard format set out under international guidance. This graph shows how GlaxoSmithKline’s policy on sharing CSRs compares to others’.

Measures

We assess every company’s policy by reading through their commitments, and checking off whether they meet a range of criteria. We send every company our assessment, then apply a scoring system to each element. You can read about the scoring system here. You can read our detailed assessment of whether GlaxoSmithKline meets each of our criteria, below.

csrs_1: Do they have a policy on sharing Clinical Study Reports (CSRs) at all?
Yes
csrs_2: Do they commit to share CSRs?
Yes
csrs_3: Is access to CSRs on request only, rather than prospectively posting CSRs online? (If only on request, summarise in comment how onerous the request process is, e.g. is it the same high level of workload as for a full IPD request, with extensive review of request and requesters; or is this just being used to prioritise which CSRs to share by the company?)
No / proactively posting.
csrs_4: Spare coding space for CSR issues [not currently used].
csrs_5: Does this commitment to sharing CSRs include trials on unlicensed treatments?
No
csrs_6: Does this commitment to sharing CSRs include unlicensed uses of licensed treatments? (In the absence of a clear commitment either way: if the company does not share CSRs on unlicensed treatments, the answer on unlicensed uses of licensed treatments is assumed to also be "no"; if the company has a clear theme of "all" trials throughout their policy, then the answer is assumed to be "yes").
Unclear
csrs_7: Does the policy commit to sharing synopses only? (This means: are synopses the only part of the CSR ever made available? If actual CSRs are available on request, and the synopses are routinely published, then this field is coded "no", and the CSR sharing policy is coded as for their policy on sharing CSRs proper).
No
csrs_8: Does their current policy cover CSRs of past trials?
Yes
csrs_9: Is access to CSRs on request only, rather than prospectively posting CSRs online? (If only on request, summarise in comment how onerous the request process is, e.g. is it the same high level of workload as for a full IPD request, with extensive review of request and requesters; or is this just being used to prioritise which CSRs to share by the company?)
No
csrs_10: Does this commitment to sharing past CSRs include trials on unlicensed treatments?
No
csrs_11: Does this commitment to sharing past CSRs include unlicensed uses of licensed treatments? (In the absence of a clear commitment either way: if the company does not share CSRs on unlicensed treatments, the answer on unlicensed uses of licensed treatments is assumed to also be "no"; if the company has a clear theme of "all" trials throughout their policy, then the answer is assumed to be "yes").
No
csrs_12: Does the policy commit to sharing synopses only? (This means: are synopses the only part of the CSR ever made available? If actual CSRs are available on request, and the synopses are routinely published, then this field is coded "no", and the CSR sharing policy is coded as for their policy on sharing CSRs proper).
No
csrs_13: From what date does this policy on sharing CSRs apply? (Give explanation of how year is used eg "trials initiated after 2001" or "trials used in regulatory approval since 2003" etc, or if not other details given then "200x NOS" [not otherwise specified]).Note after the second "/" this date is normalised, as per previous date column. "Filed with regulator" assumed to be same as "completed".
2000 nos / 2000
csrs_14: Details on additional exclusions and redactions regarding CSR sharing may be posted here, but the definitive source is the full text of the policy.
We have committed to publish the csrs for all new studies on our medicines – both medicines that are approved by regulators and ones that are terminated from development. these will now be included alongside the other study information on our online register. in addition, we are also publishing historic csrs for clinical outcomes trials for our approved medicines dating back to when gsk was formed in 2000. this is a significant volume of studies and so we have put in place a dedicated team for this programme of work, to retrieve and examine the historic csrs and remove any confidential patient information. this will happen over the next few years and we are posting the reports in a step-wise manner, starting with csrs for our most commonly prescribed medicines.

1 out of 46

Individual patient data

Why Individual Patient Data is important

Individual Patient Data (IPD) is the raw data collected during a clinical trial, with detailed information on each individual participant. It presents significant opportunities for research. For example, access to IPD can allow third parties to verify that trialists analysed their data correctly and fairly. It helps researchers combine data from lots of trials for more accurate comparisons of treatments. It helps see if a treatment is particularly effective, or unhelpful, in a subset of patients. It also allows entirely new hypotheses to be explored in existing data, helping to devise or refine new treatments. However it also presents a risk: individual patients can often be re-identified, and their privacy compromised, even with partially anonymised data. Because of this, IPD is not generally posted in public, but shared through various controlled access mechanisms, as with other forms of electronic health record data already used by medical researchers. This graph shows how GlaxoSmithKline’s policy on sharing IPD compares to others’.

Measures

We assess every company’s policy by reading through their commitments, and checking off whether they meet a range of criteria. We send every company our assessment, then apply a scoring system to each element. You can read about the scoring system here. You can read our detailed assessment of whether GlaxoSmithKline meets each of our criteria, below.

ipd_1: Do they have a policy to make individual patient data (IPD) from clinical trials available on request?
Yes
ipd_2: From what date does this policy on sharing IPD apply? (Give explanation of how year is used eg "trials initiated after 2001" or "trials used in regulatory approval since 2003" etc, or if not other details given then "200x NOS" [not otherwise specified]). Note after the second "/" this date is normalised, as per previous date column. "Filed with regulator" assumed to be same as "completed".
Start in or after 2013, but note could have been 2000 if willing to count "global trials" / 2014
ipd_3: Does this commitment to sharing IPD include trials on unlicensed treatments?
Yes
ipd_4: Does this commitment to sharing IPD include unlicensed uses of licensed treatments? (In the absence of a clear commitment either way: if the company does not share IPD on unlicensed treatments, the answer on unlicensed uses of licensed treatments is assumed to also be "no"; if the company has a clear theme of "all" trials throughout their policy, then the answer is assumed to be "yes").
Yes
ipd_5: Does the policy include phase 4 trials? If they say "all" then this is assumed to include phase 4.
Yes
ipd_6: Are there any additional exclusions? (For IPD sharing the answer is almost always "yes").
Yes / 6 months after publication. (unclear what they do if not published)
ipd_7: Do they say they consider requests for IPD on additional trials not explicitly covered by their policy?
Yes
ipd_8: Details on additional exclusions and redactions regarding IPD sharing may be posted here, but the definitive source is the full text of the policy, and a structured review of all restrictions in IPD sharing policies is beyond the scope of this audit.
Clinical studies where data labels and/or supporting documents are not in english. clinical studies of rare diseases. this is because anonymisation of these data is more difficult to achieve. for these studies gsk will assess the feasibility of anonymisation as part of the review of enquiries. clinical studies of gsk consumer healthcare products. this is because gsk anticipate more interest from researchers in accessing data from studies of pharmaceutical medicines and vaccines. when patients agreed to take part in gsk clinical studies they gave permission (through informed consent) to use their data to study the medicine or disease gsk were researching. further research must therefore study the medicine or disease that was researched in the original studies. for studies gsk conduct from 2013, patients will be asked to give permission for broader research so other research may be possible with data from these studies.whether the studies have been published or accepted for publication and whether they researched terminated or authorised medicines (for approved uses). it is gsk policy to provide access to patient level data after authorisation or termination of the medicine and acceptance of the study for publication in the scientific literature. whether gsk are able to provide the requested data. for example, the majority of non-interventional (or observational) studies use data from third party databases under licence agreements which prevent gsk from providing access to the data. researchers can access data directly from these third party databases under similar agreements. whether gsk have the legal authority to provide the data. for example, gsk may not have the legal authority because the medicine has been out-licensed to another company. whether gsk consider it feasible to anonymise the data without compromising the privacy and confidentiality of research participants. for example, anonymisation of data from studies of rare diseases is more difficult to achieve and will be reviewed on a case-by-case basis. whether gsk consider that there are any practical constraints to providing access to the data. for example, there may be issues related to the size of databases from genetic studies. the resources (costs) for gsk to retrieve data and documents from repositories and archives, anonymise data, and redact personally identifiable information from relevant documents. in some cases, particularly for older studies, the costs could be considerable and gsk may turn down requests on this basis.
ipd_9: Changes since 2015? Notes.