As part of our need-finding research, we evaluate problems across two axes: 

1. Frequency: Does this happen everyday, or once a year?

2. Pain: Does the user care enough to want it solved?

We discovered one such problem when a doctor approached us with a data management problem for a multi-year oncology trial spanning 1,080 patients across 6 sites over 5 years, with roughly 30 visits per patient.

Problem overview

The trial involved a combination of eight chemotherapy drugs that clinicians routinely prescribe. However, within the context of a clinical trial, they have to deviate from their established standard of care practices and instead follow detailed trial protocols specific to each case

In addition, they are responsible for maintaining treatment logs, completing detailed case report forms, tracking adverse events, and eventually reviewing the data  for analysis. This is layered on top of already demanding  workloads in high-morbidity settings.

It seemed to us that managing such a trial using paper-based systems would be highly impractical. We therefore dug deeper to understand the gaps.

Trial data capture and processing workflows are largely manual

Clinical trial documentation workflows in most academic settings in India remain largely manual. Detailed case record forms (CRFs) are required to be filled from source documents - such as handwritten clinical notes and diagnostic reports. In the trial that we dealt with, each patient had, on an average, of over 100 pages of handwritten records. These had to first be transcribed into the electronic medical record to remain as “source notes” for the study. Using these, the study team completed a detailed case record form, which also spanned 70+ pages. Finally, these paper forms had to be transcribed into a digital, analysable format such as an Excel sheet.  

Transcription error rates in such settings have been reported to be approximately 9–10% per patient encounter, including incorrect dosages (e.g., 0.5mg vs 5mg), missing identifiers, and inaccurate dates [1]. These issues are further compounded by the fact that data entry is often performed by personnel without formal clinical training, increasing the likelihood of misinterpretation. Additionally, repeated transcription of the same data across formats introduced multiple opportunities for error.

Manual Adverse Event Reporting Impacts Data Quality and Timeliness

Timeliness of documentation also presents a major challenge. Regulatory requirements mandate that serious adverse events (SAEs) be reported within 24 hours to ethics committees and regulatory authorities. In practice, this requires manually reviewing patient records, correlating symptoms with treatments, tracking vital trends, applying CTCAE standards to identify adverse events and documenting each event in detail. In the setting we observed, there were around 30 such events per month per patient, with each case taking roughly 30 minutes to process from paper files.

Given limited resources and competing clinical demands, the quality and timeliness of this data are often impacted. Adverse drug reaction reporting rates in India are estimated to be below 1%, compared to a global average of around 5% [2]. Delays and underreporting in such workflows directly impact both patient safety and regulatory compliance.

How Foster solves this

Foster addresses this by rethinking the documentation workflow from the ground up. Study teams can scan and upload patient records into structured folders, creating a digital source of truth from the start. They can configure forms aligned with their study protocols, and Foster automatically extracts and populates the case report forms. Instead of manually entering data, teams review and validate outputs. The final data is already structured and ready for analysis, eliminating the need for repeated transcription.

Our goal is to reduce documentation errors and delays in adverse event reporting by at least 50%, and make it feasible to run high-quality investigator-initiated studies at scale.