Impact of the COVID-19 Pandemic and Collection/Analysis of Laboratory Data in Ongoing Clinical Trials


INTRODUCTION

Since late 2019/early 2020, the global COVID-19 pandemic has impacted all aspects of daily life. Ongoing clinical trials have been impacted in multiple ways, including but not limited to delays in screening and enrolment of subjects into trials, challenges with maintaining the trial schedule for ongoing subjects due to temporary site closures, lack of continuous drug supply, and subjects experiencing new and unexpected adverse events. The safety of clinical trial participants must be ensured despite these challenges. 

Many global health authorities have issued guidance documents with advice and instructions to sponsor companies and clinical investigative sites on how to handle disruptions related to the COVID-19 pandemic. These guidance documents are published on the official websites of the US Food and Drug Administration (FDA) [1], the European Medicines Agency (EMA) [2] and many other countries (including individual member states within the EU). This PHUSE paper is meant to supplement the information in these guidance documents and in no way is meant to replace or contradict instructions from the health authorities. 

This paper focuses on points to consider and some general recommendations for handling of laboratory data from ongoing clinical trials impacted by the COVID-19 pandemic in the areas of data collection, data monitoring, data transformation, and data analysis. It is important that the reader consider the specific circumstances of each impacted trial including subject population, indication being studied, length of trial, number of subjects, and local restrictions. Many sponsor companies have formed COVID-19 task force teams, which are publishing documents to supplement company SOPs and best practices.   The reader is advised to consult with your specific sponsor company’s rulebooks and instruction documents. The collection and analysis of data in trials for the treatment of COVID-19 is out of scope of this paper.


DATA COLLECTION

Due to the temporary closure of many clinical trial sites and/or restriction of activities in many hospitals, ongoing trial visits may be conducted in ways other than specified per protocol. Visits may be conducted as televisits via phone or video rather than as physical in-person visits. While this is beneficial for checking on subject status and to document changes in adverse events and concomitant medications, televisits do not easily support the collection of laboratory data. If warranted and feasible, the investigator may instruct the subject to visit a local laboratory not affiliated with the trial to undergo sample collection. The investigator should specify for the subject the list of minimum laboratory parameters and any additional parameters of special interest to be reported.  

The sponsor company must then determine how to accommodate the data from these additional sources. It may be helpful to consider how often the collection of lab data from non-trial associated laboratories is occurring and the volume of data collected this way in order to determine how best to handle this data. There are two likely scenarios:

Scenario Description Points to Consider
Scenario A: Do not include the individual results from the non-trial associated laboratory in the trial database. Use this data to support safety monitoring only. Report any unusual findings as adverse events.  One risk with this scenario is the related individual laboratory results not being stored in the trial database in such a way that they can be included in any analyses or subject profile. 
Scenario B: Include the individual results from the non-trial associated laboratory in the trial database. 
  • In addition to laboratory results, the laboratory name, sample type, normal ranges, method, fasting status, LOINC code, and other supporting data must also be entered into the trial database in addition to sample collection and results reporting. These data are necessary to ensure proper conversion to standard units, if required.
  • Consider how the trial database is set up to record this additional data. Some trial databases may already be able to accommodate the reporting of local laboratory results. Other trial databases may not be set up to accommodate local lab results, and a database update would be required. Consideration should be given to the amount of effort required to perform the database update.
  • Consider how additional parameters not specified by protocol but reported would be handled. Will these be included in the trial database?
  • Consider if and how the data will be used in analyses.


While the investigator may request the subject visit a local laboratory not affiliated with the trial to undergo sample collection, the sponsor may also request that all subjects undergo such testing. Before recommending this, the sponsor should consider:

  • the additional burden on the subject to undergo a sample collection at an unplanned location
  • the additional burden on the trial site to enter this data (normal ranges, methods, LOINC codes, sample collection information, results, etc.)
  • the additional burden on the sponsor company to perform any database updates and additional data reconciliation and data cleaning activities related to the collection and use of this data, as outlined in scenario B above.

Alternatively, if the site or sponsor misses the opportunity to collect such data, it may result in additional burden on the sponsor company and regulatory reviewer. If lab summaries are somehow incomplete, it may be more difficult to interpret analyses.

It may be useful to consider a hybrid approach where there is a conscious decision by the medical team at the sponsor company regarding which visits can be missed and for which visits and/or specific laboratory parameters an attempt should be made to collect data via a local laboratory not affiliated with the trial.

If trial sites remain open, it may also be the case that subject visits may occur as scheduled, and sample collection can take place as expected but the samples do not arrive in time at a central laboratory due to shipping delays. In these cases, the sample collection should be documented with reason not done, as per standard procedure.

If any enrolled subjects in the trial are diagnosed as COVID-19 positive, this must be reported as an adverse event and coded with the specific MedDRA defined code. Any additional laboratory data, including COVID-19 test results and/or COVID-19 antibody test results, should be handled as determined within the trial following scenario A or scenario B.  

Any additional approaches to collection and reporting of laboratory data which deviate from the original protocol should be documented in the study data management plan.  

 

DATA MONITORING

Monitoring efforts should focus on the following key areas:

Scenario Description Points to Consider
Missed visits or visits conducted via televisit, not allowing for laboratory sample collection. This scenario should likely be documented as a protocol deviation.
Scenario A — Alternative methods used to collect laboratory data and data IS NOT reported in trial database. Ensure copies of laboratory results are available in the subject’s source documentation. Ensure consistency between any adverse event entries and supporting laboratory result documentation.
Scenario B — Alternative methods used to collect laboratory data and data IS reported in trial database.

Ensure all required data are collected. This may include:

  • items to support conversion to standard units (sample type, normal ranges, method, age, gender, LOINC code, etc.)
  • sample collection details (date and time, laboratory, fasting status, etc.)
  • all information required to support data reconciliation and data cleaning
  • reason certain parameters not reported (if incomplete results reported).
Samples collected out of window.

This scenario should likely be documented as a protocol deviation.


Updates to the study monitoring plan should be made to outline any additional approaches to collection and reporting of laboratory data which deviate from the original protocol. 


DATA TRANSFORMATION

Data transformation rules and/or programs from source data to SDTM and/or ADaM may require revision to accommodate any data collected and reported under Scenario B.  

During mapping of source data to SDTM, conversion to standard units may be required to ensure consistency. Additional rules for identification of protocol deviations may be required. 

During creation of ADaM data, data handling rules for missing data or data collected out of the planned visit windows should be considered. Will data imputation be performed? If yes, is there any impact on any defined data imputation rules? Will visit windows be defined (if not already specified) or widened (if already specified)? How will partial results be handled (some parameters reported, other parameters missed at the same timepoint)?

Updates to the data transformation specifications, Define-XML documentation, study and/or analysis reviewer’s guides, the statistical analysis plan, and/or data monitoring plan may be required. 



DATA ANALYSIS

The general recommendation of this paper is to keep with the planned, pre-specified analyses as much as possible. Most sponsors likely have planned analyses identical or similar to those provided in PHUSE (2013) [4] and PHUSE (2015) [5]. Consider the balance across treatment groups with respect to number of missed visits, changed visits, alternative data collection methods, protocol deviations, etc. If the impact from modifications required in the trial conduct is minimal and/or is balanced across treatment groups, the pre-specified analyses can likely proceed as planned. In these cases, it is recommended to not overcomplicate the clinical study report with additional “clutter” and sensitivity analyses that won’t add value to the understanding of the benefit/risk of the investigational product. 

When Scenario B is implemented in a trial, consider the handling of additional laboratory parameters. It is recommended to analyse only those analytes specified in the protocol. If additional analytes or additional timepoints are collected, they can be presented in subject listings and subject profiles as unscheduled assessments, but they do not have to be analysed. When creating summaries of the observed data and/or changes from baseline, the study team will need to decide which laboratory measurements can be combined. For some analytes, directly combining the data may not be appropriate. Alternatively, study teams can choose a different analytical approach that allows for combining laboratory measurements from different laboratories. A frequently used technique is to report the data as a percent above/below the normal limit. Alternatively, a normalisation method [7, 8] can be used to combine local and central labs in the analysis.

Special consideration should be given to any laboratory parameters that are efficacy assessments. Here, a trial-specific solution is likely required after discussion within the project or study team. If the parameters are relatively “standard”, such as fasting glucose for a diabetes trial, it may be that these can be easily collected by a local laboratory not affiliated with the trial or collected by the investigative site yet out of window. In other studies, the efficacy parameters may not be so “standard”, in which case consideration should be made regarding how to deal with missing data.   

It is anticipated that a good number of planned visits may be missed completely or performed out of window. Changes to the data handling/data imputation/visit windowing rules should be considered if many samples are being thrown out of the analysis. In general, it is likely better to use available results than lose data which are not conforming to such rules. If many samples are collected out of window, it should be considered to change table presentations from by visit presentations and to change to minimum or maximum values. This approach is described in more detail in PHUSE (2013) [4].

For long-term studies, the reader may consider establishing a specific date window when the COVID-19 pandemic had the most impact on the trial. For this time period, specific algorithms may be applied for this time window only. The pre-planned algorithms may be applied to any data collected pre or post the defined time window of this COVID-19 pandemic. Analyses may also be performed by comparing data of subjects impacted by COVID-19 versus data of subjects not impacted by COVID-19. This may be most applicable to lab analytes of special interest, and not all safety labs. 

It is the general recommendation of this paper not to plan on any specific subgroup analyses for signal detection related to any changes in data collection due to the COVID-19 pandemic. If the observation time between treatments is similar due to any disruptions, then a key aspect for signal detection of safety issues is met and additional analyses should not be required. If there was a severe disruption in collection of data or an imbalance across treatment groups, then the reader may consider whether any additional subgroup analyses are appropriate.

If any enrolled subject is diagnosed as COVID-19 positive, the recommendation of this paper is not to perform any additional special analyses unless determined on a trial-specific level. Any COVID-19 related information should be included as part of the subject profile.

Any changes to the pre-specified analyses, additional analyses, new or modified table displays, etc. should be documented in an amendment to the statistical analysis plan or in a supplemental statistical analysis plan. 



CONCLUSION

Changes to data collection, data monitoring, data transformation and data analysis of laboratory data in ongoing clinical trials due to the COVID-19 pandemic need to be discussed at the study level for each impacted trial. In general, it is advisable to follow the pre-planned, pre-specified data collection and analyses for the trial, although special consideration may be made depending on subject population, indication, length of trial, and other factors. Ensuring appropriate safety monitoring of trial subjects is of utmost importance and therefore considerations of alternative data collection and review as outlined in this paper should be considered.



REFERENCES

[1] FDA Guidance on Conduct of Clinical Trials of Medical Products during COVID-19 Public Health Emergency Guidance for Industry, Investigators, and Institutional Review Boards – March 2020, Updated on May 14, 2020.

https://www.fda.gov/media/136238/download

[2] Guidance on the Management of Clinical Trials during the COVID-19 (Coronavirus) Pandemic – Version 3 (28 April 2020).

https://ec.europa.eu/health/sites/health/files/files/eudralex/vol-10/guidanceclinicaltrials_covid19_en.pdf

[3] Clinical Trial Drug Safety Assessment for Studies and Submissions Impacted by COVID-19 – Mary Nilsson, Brenda Crowe, Greg Anglin, Greg Ball, Melvin Munsaka, Seta Shahin, Wei Wang.

https://arxiv.org/abs/2006.05502

[4] PHUSE (2013): Analysis and Displays Associated with Measures of Central Tendency – Focus on Vital Sign, Electrocardiogram, and Laboratory Analyte Measurements in Phase 2-4 Clinical Trials and Integrated Submission Documents. 

https://www.phuse.eu/documents//working-groups/deliverables/analyses-displays-associated-with-measures-of-central-tendancy-version-10-10-oct-13-11816.pdf

[5] PHUSE (2015): Analysis and Displays Associated with Outliers or Shifts from Normal to Abnormal: Focus on Vital Sign, Electrocardiogram, and Laboratory Analyte Measurements in Phase 2-4 Clinical Trials and Integrated Submission Documents.

https://www.phuse.eu/documents//working-groups/deliverables/analyses-displays-associated-with-outliers-or-shifts-from-normal-to-abnormal-version-10-10-sept-15-11815.pdf

[6] Guidance for Ongoing Studies Disrupted by the COVID-19 Pandemic.

http://www.cdisc.org/system/files/members/standard/ta/COVID-19/Guidance%2520for%2520Ongoing%2520Studies%2520Disrupted%2520by%2520COVID-19.pdf

[7] From Local Laboratory to Standardisation and beyond: Applying a common grading system.

Angelo Tinazzi, Irene Corradino, Enrica Paschetto, Sonia Colombini. SENDO-Tech S.r.l., Milan, Italy

https://www.lexjansen.com/phuse/2007/dm/DM01.pdf 

[8] Karvanen, J. (2003). The Statistical Basis of Laboratory Data Normalization. Drug Information Journal: DIJ/Drug Information Association, 37, 101—107.

 

Posted by on

Tags: COVID-19

Categories: Working Groups News

Related Blogs

Add Your Comments

Thinking of joining PHUSE?

Already a member but not sure how you can benefit?

PHUSE is an expanding, global society with a global membership of clinical data scientists. It requires a large pool of resources to help with its running, and so there are many opportunities for members to become involved. Whether it's chairing a conference, presenting at an event, leading a working group or contributing to the quarterly online newsletter, we are always keen to hear from volunteers.

Find Out More