The process of safety narrative writing has changed drastically over the years. But despite all these changes, the goal remains to deliver a high volume of narratives describing key events, the details of which are important for drug safety reporting. As technology is evolving, the process of developing patient safety narratives is also evolving.
The Role of Safety Narratives in Drug Development
Safety narratives support the evaluation of drugs’ safety profile by summarizing adverse events (AEs) and relevant information. This summary helps researchers and other parties, including pharmacovigilance, regulatory affairs, the clinical team, the data safety monitoring board, and regulatory authorities, identify safety issues that may require further investigation. Further investigation may require additional clinical trials or narratives being written specifically for certain events. This information helps researchers and regulatory authorities characterize the relationship between the drug and the AE.
Per ICH E3, a safety narrative should describe deaths, serious adverse events (SAEs), nonserious AEs leading to study discontinuation, and AEs of special interest (which depend on the indication, disease under study, and study type). The medical writer uses a variety of sources, including clinical database listings, patient profiles, Council for International Organizations of Medical Sciences (CIOMS) forms, MedWatch forms, and case report forms to develop the safety narrative.
How Narrative Writing has Changed
Generally, the narrative writing process involves preparing a template (which maps where each piece of information is pulled from in the data), preparing sample narratives, and then once the samples are approved, the medical writer drafts the narrative from patient data. Then another medical writer conducts peer review, followed by clinical and safety review by the Sponsor. The narratives are revised as needed and undergo quality control (QC) review, approval, and finally, publishing as part of a clinical study report or other dossier.
When I started medical writing approximately 15 years ago, all narratives (and all pieces of narratives) were done manually. We had to search through thousands of pages of listings by patient number and then enter each piece of data manually into the narrative template, which was also created and mapped manually. This left potential for error as we typed information into the narrative when looking between documents. In addition, it took a very long time! For example, with this sample sentence shown below from a narrative, each of the bolded pieces of information would have to be located and pulled manually from the listings. It was typical to spend 4 to 8 hours on each complex narrative and only complete 1 to 2 narratives per day.
Subject 1234, a 42-year-old White female, was enrolled in Study 98765 and started treatment with Drug X once a week and Drug Y once every 2 weeks on 04 March 2021 (Day 1)
Nowadays, we have programmed patient profiles and programmed portions of narratives, which saves us so much time. We don’t have to comb through thousands of pages looking for just the patient we are writing about. Instead, each of these bolded data pieces are pulled programmatically into the narrative. We can focus on what happened to the patient and what is relevant to the events. The only pieces that need to be added manually are items from the CIOMS and medical interpretation. This technology makes the narrative more accurate, as we are not manually pulling data from one document to another. And it allows the medical writers to focus on what is important, rather than spending a lot of time adding pieces manually that we can do programmatically instead. Because we save so much time by using these programmed options, we are less likely to have to draft narratives using pre‑database lock data, which are often dirty. This saves time overall, as the narratives are drafted using final clean data and don’t have to be updated, and there is less chance for error.
In the “old days” of completing narratives without programming help, we typically had a small header at the top, where we manually entered the patient number and maybe the events triggering the narrative, and then the rest was in paragraph form. We’d write a long paragraph giving the patient’s demographics and medical history, followed by a second long paragraph summarizing the concomitant medications, then more long paragraphs summarizing other background information, previous events, and the events of interest. It was cumbersome and wordy.
Nowadays, the programming allows us to start with a patient profile or programmed header table that has the complete medical history, concomitant medications, baseline information, etc., already presented. Then the text is a clear and concise “story” of what happened. It’s easier to read and sort through. And the shorter text is also easier to QC! There is more of a focus now on being concise and only including relevant information, rather than just dumping everything into the text.
The Future of Narrative Writing
The use of technology to assist in programming narratives, and the efficiencies subsequently produced, lead to companies being able to easily respond to a regulatory authority’s request for narratives. Our software, Synchrogenix™ Writer, contains a narrative builder feature that allows us to program narratives quickly and efficiently. Turnaround times are faster because of the time saved. And the narratives produced are higher quality as there is less room for error with a more focused team and output and a more streamlined process. In the future, narrative writing will become even more streamlined because of technology‑enabled authoring. Technology performs most of the time-consuming work for the writing team. But this technology cannot eliminate the need for well-educated, experienced medical writers, who can sort through relevant information and make sure the narratives focus on the important details.