Top Tips for CSR Safety Narratives

Written by Dr Steph Jones

CSRs (Completed Study Reports)

Whether you are working with an external service provider or creating them in-house the safety narratives always cause a great deal of work right at the end of the study. This can be the busiest period when the team needs to focus on generating key messages for the completed study report (CSR) and potentially a marketing authorisation application. However, this is often when resources have been pulled onto other studies.

The quality of the CSR narratives needs to be high for several reasons. For example these are used by regulatory reviewers to provide context for adverse events when they see a concern in the CSR. This means they need to present the data in a way that is easily accessible by the reader and any plausible alternative causal factors potentially contributing to the events need to be clear rather than diluted by reams of irrelevant description in the written summary. Poor quality CSR narratives can reflect badly on the rest of the submission when sending as part of the application for marketing authorisation.

Read on to find some tips on how to ease your workload at this busy time and make the production of CSR narratives as smooth as it can be.

1) Early preparation

Narratives are a requirement in every completed study report (CSR) therefore the production of this critical element could be factored into the planning right from the start of the study. This will allow front-loading of as much of the work as possible and will also facilitate the optimal allocation of resources throughout the timeline of the study.

2) Decide on narrative categories and discuss with agencies

There are standard categories of patients for whom a narrative is expected by the regulatory agencies. Per ICH E3 section 12.3.2 (Structure and Content of CSR) safety narratives are: “brief narratives describing each death, each other serious adverse event, and those of the other significant adverse events that are judged to be of special interest because of clinical importance”. However, there could be variations from the usual narrative categories depending on the needs of specific clinical trials. In those cases it is important to discuss this with the agencies in advance as they may not always agree, particularly where the categories are reduced from the expected minimum. It is also important to clearly define what is meant by “Adverse event of special interest” so that patients with these events can be easily identified.

3) Templated tables

Wherever possible a standard template should be used in the tabulated part of the narrative. There are components of every narrative such as demographics, medical history, concomitant medications, exposure to the investigational product(s) and adverse events that are collected in a standard way across different trials. Once the clinical study protocol (CSP) is approved any variations from this standard should be easy to spot and therefore a study-specific template can begin to be created. There is a great deal of programming behind these tables and using a standardised approach across studies can minimise the amount of “re-work” and streamline the resources to focus only on those parts where there are variations.

4) Templated written summary

There is a written summary at the end of each narrative to explain the relevant features of the patient journey and provide context to the adverse events. There are many components that can be presented in a standardised way across studies, including demographics, medical / surgical history and the name of the trial. In many cases it is also possible to template exposure and the presentation of adverse event data. Building a standard approach to reporting these can speed up the writing of narratives whichever process is used to generate them.

5) Data set preparation for the completed study report

These narratives cannot be generated from the raw data in the electronic data capture (EDC) system as the programming is designed for the final analysis data set. The final dataset therefore needs to be generated before narratives can be produced. This is why narratives cannot be created until after database lock (DBL). It could be possible to generate a draft set of narratives at the same time as a “Dry run” if there is a blinded delivery for review (BDR) created to check the programmed outputs for the rest of the CSR.

6) Draft narratives review for the completed study report 

If drafts are created during a BDR it is important to remember that the data cut off (DCO) will normally be many weeks or even months earlier than the current data in the EDC. This is usually at a time of intense data review activity with queries being addressed by sites and data being updated, which means the data in these narratives may be out of date. The main purpose of reviewing the drafts is therefore to check the format is correct and that the programming pulls data into the correct fields. It is also important to remember that these will remain blinded until after DBL, so any changes made to the template at this time need to consider the impact of amending blinded data to unblinded in the final version.

It can sometimes be possible to create a written summary in advance of DBL, in particular for patients who completed the study or who are no longer contributing safety data, or for patients who are likely to contribute minimal data by the time of the DBL. This can front-load some of the work to reduce the workload after DBL as many of them will need little or no adjustment once the final dataset is generated. However they will need cross-checking to make sure.

7) Preparing for final CSR narratives for the completed study report

Listings can provide the team with a fairly accurate number of predicted narratives that will be required post-DBL. Once you know the scale of the work you can predict how many people will be needed to support this activity. If people are brought in from other study teams or external service providers you need to factor in sufficient time for training and orientation on your particular study, in addition to time for any contractual interactions. If external service providers are used the internal study team needs to allow time to review iterations of the written summaries and ensure they are happy to approve the final versions.

8) Delivery of final CSR narratives: teamwork and training 

It’s “All hands on deck” when the narratives are generated. Normally this happens in batches and the medical team are provided with a large number to review so being prepared can make this process more slick. Making sure everyone is aligned on how to review is key to producing high quality narratives, reducing re-work and in particular to eliminate duplication of effort. Training and kick-off meetings should be scheduled before the work begins and then regular meetings should be held during the review process between those involved so that everyone remains aligned.

9) Delivery of final CSR narratives: coordination

A secure document storage area will often be the ideal solution as this means the narratives can be sent for review, worked on and different versions filed ready for next steps. This minimises the risk of incorrect versions being sent by email and avoids any risk with sensitive data being sent between different email systems. Naming the folders appropriately and ensuring everyone understands how to use them is critical. Appropriate training and dividing the workload is key to a slick process. Ideally one person should be nominated to coordinate the reviews and maintain a tracking system. The external service provider would normally do this for their part but the study team members need to coordinate internal activities if several people are working on narratives at the same time.

10) Final Quality Control

It is assumed that the Quality Control (QC) of tabulated portions has already been performed during the draft review stage, but if that stage is missed earlier it is important to now ensure the data in the tables represents the data in the EDC. At this stage the data should be consistent with the final dataset as it was generated after DBL and will not have changed so the issues with reviewing drafts with a different DCO do not apply.

Normally at the stage of reviewing final narratives the QC will fulfil several purposes: check whether the written summary reflects the relevant data in the tabulated portion; ensure the summary makes sense to the reader and is not confusing; determine whether the summary provides sufficient context to the events.


Narratives are a mandatory part of every CSR. They help regulatory agencies understand the patient journey for specific clinical trial participants with events of interest. The quality of these important narratives will reflect on the overall quality of a submission. Preparing well and having a coordinated, standardised approach can help you to produce the required number of narratives efficiently and to a sufficiently high quality.

Get in touch if you would like to know more about how Pharmora can support with your narratives.


Read More Pharmora Blog Posts Below

Men’s Health Week

Men’s Health Week

This week is Men’s Health Week (10th – 16th June 2024)! Please read to learn more about Men’s Health, research and health tips and current medical advancements.

read more
BioNow Careers Fair

BioNow Careers Fair

Pharmora were invited to present to the attendees of the BioNow Careers Fair. Find out how it went in this post from the presenter herself.

read more