A source document in clinical research is any original record where a patient’s data is first captured. That could be a hospital chart, a lab printout, a signed consent form, an imaging scan, a pharmacy dispensing log, or even a patient’s own diary. The concept is straightforward, but source documents sit at the foundation of every clinical trial’s credibility. Without them, there is no way to confirm that a drug actually worked, that a patient actually consented, or that the data submitted to a regulator reflects what happened in the exam room.
Why the “Original” Part Matters
The defining feature of a source document is that it is the first place information gets recorded. If a nurse measures your blood pressure and writes it on a worksheet before typing it into a database, that worksheet is the source document. If a lab machine spits out a printout with your blood count, that printout is the source document. The electronic database or case report form where the data eventually lands is a copy, and copies are only trustworthy to the extent that someone can trace them back to the original.
This chain of traceability is the entire reason source documents exist as a regulatory concept. Agencies like the U.S. Food and Drug Administration and the European Medicines Agency require that every data point in a trial can be verified against its source. If a sponsor claims their experimental cancer drug shrank tumors by 40%, a regulator needs to be able to pull up the original radiology reports, the pathology slides, and the physician’s notes to confirm that finding. Source documents are the receipts.
What Qualifies as a Source Document
There is no closed list. Anything that constitutes an original recording of a trial-related observation or activity can be a source document. In practice, the most common ones include:
- Medical records: physician progress notes, nursing assessments, discharge summaries, and consultation reports from the hospital or clinic where the patient is being treated.
- Lab reports: original printouts or electronic outputs from hematology, chemistry, urinalysis, and other laboratory panels.
- Imaging results: radiology reports, MRI or CT scan images, X-rays, and echocardiograms.
- Pharmacy records: dispensing logs that show which patient received which drug, at what dose, and when.
- Informed consent forms: the signed, dated document proving a participant agreed to join the trial after understanding its risks.
- Patient diaries: paper or electronic logs where participants record symptoms, medication use, or quality-of-life information between clinic visits.
- Device outputs: ECG tracings, spirometry readings, blood glucose meter downloads, and data from wearable sensors.
The breadth of this list is part of what makes source documentation challenging. A single patient visit in a complex trial might generate a dozen different source documents spread across paper charts, electronic health records, central lab systems, and imaging archives. Keeping all of that organized, complete, and accessible for years after the trial ends is a real operational burden on research sites.
The Scale of Documentation Failures
If source documentation were easy, regulators would not spend so much time flagging it. But it is consistently one of the top problems found during inspections. One review of documentation practices at investigator sites noted that the most common inspection finding is a lack of reliable, accurate, and adequate source documentation, and that the same problem is also the most frequent issue uncovered during sponsor audits.1Europe PMC. Good documentation practice in clinical research
The FDA reinforces this picture through its warning letters to clinical investigators. A study examining those letters found that failure to maintain accurate and adequate case histories was the second most common violation, appearing in roughly 58% of warning letters issued to investigators.2Indian Journal of Medical Ethics. A study of warning letters issued to clinical investigators and institutional review boards by the United States Food and Drug Administration A separate analysis of FDA warning letters from 2010 to 2014 similarly found that failing to maintain accurate case histories was the single most common violation cited, at about 11% of all coded violations.3PubMed. A Retrospective Analysis of Clinical Research Misconduct Using FDA-Issued Warning Letters and Clinical Investigator Inspection List From 2010 to 2014 The two studies use different denominators and time periods, which explains the different percentages, but the message is the same: documentation lapses are among the most reliably caught problems in clinical research.
Beyond the U.S., inspections at sites outside the country tell a similar story. An analysis of the FDA’s Clinical Investigator Inspection List for non-U.S. sites found that documentation and reporting deficiencies were flagged in about 24% of inspections that received a “voluntary action indicated” classification.4Medical Research Archives. FDA Clinical Investigator Inspection List, who is Inspected and What are the Results for Countries Outside US? Documentation problems are not unique to any one country or healthcare system. They are a structural challenge of clinical trials everywhere.
Source Document Verification and Its Costs
Source document verification, or SDV, is the process where a clinical research monitor physically or remotely compares every data point in the case report form against the original source document. If the case report form says a patient’s hemoglobin was 12.3 g/dL on June 5th, the monitor pulls up the original lab report and confirms that number. This is repeated for every field, every visit, every patient. For large trials, that means tens or hundreds of thousands of individual data points.
The costs are staggering. One analysis of a phase 2 multicenter trial estimated that combined sponsor and site efforts for monitoring and SDV consumed over 61,000 person-hours at a total cost of roughly $6.1 million. That worked out to about $18 per data field verified. More striking was the cost per error actually found: approximately $4,862 for each data field error caught through the process.5Communications Medicine. Systematic data capture reduces the need for source data verification: exploratory analysis from a phase 2 multicenter randomized controlled platform trial That ratio raises fair questions about whether verifying 100% of source data is the most efficient way to ensure quality.
The alternative is targeted or risk-based monitoring, where monitors focus their verification efforts on the data points that matter most, like primary endpoints, safety data, and informed consent. A prospective study comparing targeted SDV against full SDV found that the workload savings depended heavily on how long each data query took to resolve. When queries took about 15 minutes each to handle, targeted monitoring kept the overall workload below full monitoring. But as query handling time climbed to 25 minutes or more, targeted monitoring’s workload actually exceeded full monitoring, eventually reaching a 50% increase at 45 minutes per query.6PubMed Central. Impact of a targeted monitoring on data-quality and data-management workload of randomized controlled trials: A prospective comparative study Targeted approaches trade fewer checks for more complex ones, and the net benefit depends on how smoothly those checks go.
What Good Source Documentation Looks Like
Regulators do not just want source documents to exist. They want them to be attributable, legible, contemporaneous, original, and accurate. You may see this referred to by the acronym ALCOA, a set of principles built into good clinical practice guidelines. In plain terms: you should be able to tell who recorded the data, read what they wrote, confirm they recorded it when the event happened rather than days later, verify it is the first recording, and check that the information is correct.
Meeting those standards across dozens of patients, hundreds of visits, and multiple document types is where things get messy. Research coordinators juggle the demands of patient care, regulatory paperwork, and data entry. Under time pressure, notes get abbreviated, timestamps get missed, and corrections get made without proper documentation of why the original entry changed. These are the kinds of gaps inspectors flag repeatedly.
One practical response has been the use of sponsor-provided source document templates. Rather than leaving each research site to design its own worksheets, sponsors create standardized forms that are aligned with the study protocol and the electronic case report form. When these templates are designed alongside the protocol, they reduce the risk of missing data, inconsistent entries, and excessive free-text notes that can be hard to interpret later.7Pharmaceutical Outsourcing. Should Sponsors Provide Source Document Templates? The templates essentially pre-structure the data capture so that coordinators know exactly what needs recording at each visit, in what format, and in what order.
This approach is not without controversy, though. Some investigators feel that sponsor-imposed templates can be overly rigid and do not accommodate the clinical realities at their particular site. There is also a philosophical question about whether a template designed by a sponsor truly represents the site’s “original” documentation, or whether it blurs the line between source data and the case report form. Regulatory guidance generally permits sponsor-provided templates as long as the site accepts responsibility for the data recorded in them, but the debate about best practices is ongoing.
The Move Toward Electronic Source Data
One of the biggest shifts in source documentation is the move from paper to electronic systems. In a traditional trial, a patient’s blood pressure gets written on a paper worksheet, then manually typed into an electronic case report form. That creates two records and a gap where transcription errors can creep in. Electronic source data, or eSource, aims to eliminate that gap by capturing data electronically from the start. If the blood pressure reading flows directly from a clinic’s electronic health record into the trial database, the electronic record is the source document and there is no transcription step to verify.
Software systems designed for eSource hold promise for increasing data quality while reducing the costs of data collection, monitoring, and source document verification.8PubMed Central. Defining methods to improve eSource site start-up practices The logic is straightforward: if the data is born digital and flows directly into the trial database with appropriate validation checks, there are fewer opportunities for the kinds of errors that traditional SDV is designed to catch.
But electronic systems bring their own regulatory requirements. In the United States, any electronic record that serves as a source document or is used to support a regulatory submission must comply with 21 CFR Part 11, a set of FDA rules governing electronic records and digital signatures. The requirements include maintaining data integrity, backing up data, validating all system components, safeguarding digital signatures, and maintaining a date-and-time-stamped audit trail.9PubMed. A 21 CFR Part 11 compliant graphically based electronic system for clinical research documentation The audit trail is the electronic equivalent of never using white-out on a paper chart: every change to the record must be tracked, with a record of who made it, when, and why.
Implementing systems that meet these standards is not trivial. One case study described a large clinical research organization that needed to replace its legacy interactive voice response system specifically because it did not provide audit trails compliant with 21 CFR Part 11.10The Quality Assurance Journal. 21 CFR Part 11: electronic records. An interactive voice response system. Case study for success These systems are used for tasks like randomizing patients and managing drug supply, functions that generate data with direct regulatory significance. Running them on platforms that cannot track every modification is a compliance risk that sponsors take seriously.
When Source Documents Reveal Fraud
Source documents are not just a bureaucratic requirement. They are one of the primary tools for detecting fabricated or manipulated trial data. When a researcher invents patient visits or falsifies lab results, the source documents (or their conspicuous absence) are usually what gives them away.
A compelling example comes from the European Stroke Prevention Study 2, a large multicenter trial in which one site was later confirmed to have committed fraud. Statistical monitoring methods applied after the fact identified that site as the second most atypical out of 60 participating centers. The red flags in the source data were telling: only serious adverse events had been reported, with no non-serious adverse events or other expected event types like vascular events. Laboratory results and vital signs showed unusually low variability both between and within patients. Multiple data fields had atypical proportions and suspicious patterns of missing values in areas like drug compliance and adverse event reporting.11PubMed Central. Detection of Fraud in a Clinical Trial Using Unsupervised Statistical Monitoring
That pattern makes intuitive sense. Real patients are messy. They show up with unexpected lab values, experience minor side effects, forget to take their medication sometimes, and generate data with natural variation. When source data looks too clean, too uniform, or too perfectly aligned with expectations, it is often because someone manufactured it rather than observed it. Genuine source documents carry the fingerprints of real clinical encounters: stray annotations, normal-range fluctuations, the occasional illegible handwriting. Fabricated data tends to lack those textures.
Adverse Events and the Underreporting Problem
One area where source documentation problems have especially serious consequences is adverse event reporting. When a trial participant experiences a side effect, that event needs to be captured in the source documents, assessed for severity and relatedness to the study drug, and reported up the chain. If the source documentation is incomplete or the event is simply not recorded, safety signals can be missed entirely.
An analysis of adverse event documentation in one study found an underreporting rate of roughly 98%, meaning researchers identified and documented only about 2% of the adverse events that actually occurred.12Applied Clinical Trials. Documenting Adverse Events That figure comes from a single study and should not be taken as representative of all clinical trials, but it illustrates a well-known problem in the field. Adverse events that are mild or that seem unrelated to the study drug are the ones most likely to go unrecorded. When a patient mentions a headache offhand and the coordinator does not write it down, that headache never makes it into the trial’s safety database. Across thousands of patients, those omissions add up and can obscure real safety patterns.
The fix is partly cultural and partly structural. Culturally, research teams need to treat adverse event capture with the same rigor as primary endpoint data. Structurally, source document templates that include explicit prompts for adverse event assessment at every visit can help ensure that the question at least gets asked and the answer gets recorded, even when the answer is “none.”
The Role of Quality Assurance
Source documents do not maintain themselves. The systems that keep them reliable depend on broader quality assurance programs: monitoring visits, internal audits, training for research staff, and data management processes that flag inconsistencies before they become entrenched. Good clinical practice guidelines, as defined by the International Conference on Harmonisation, frame quality assurance as essential for both participant safety and data credibility.13PubMed Central. The Role of Quality Assurance in Clinical Trials: Safeguarding Data Integrity and Compliance
In practice, quality assurance for source documents looks like a monitor sitting at a research site, pulling patient charts, and checking them line by line against the electronic database. It looks like a data manager running automated checks that flag values outside plausible ranges. It looks like an auditor reviewing whether corrections to source documents were initialed, dated, and explained. These activities are tedious, expensive, and often resented by busy clinical staff, but they are the mechanism by which the research enterprise maintains its claim that its data is trustworthy.
For anyone involved in a clinical trial, whether as a participant, a coordinator, or a physician, source documents are the thread that connects the bedside to the regulatory filing. They are not glamorous. No one goes into medicine dreaming about properly initialed corrections on a lab worksheet. But when a new therapy reaches the market, the confidence that it actually works and is reasonably safe rests in large part on whether somebody, somewhere, wrote things down correctly the first time and kept those records intact.
When Participants Are the Source
An increasingly common scenario in modern trials is patient-generated source data. When a participant fills out an electronic diary on a smartphone, or a wearable device continuously logs heart rate and activity levels, the data originates outside the clinic entirely. The participant’s home becomes the research site, and the device or app becomes the source document.
This raises questions that traditional regulatory frameworks were not designed for. Where is the “original” when data is captured by a wearable sensor, transmitted to a cloud server, processed by an algorithm, and then displayed on a dashboard? Which version is the source? Regulatory agencies have been working through these questions, generally concluding that the first fixed recording of the data constitutes the source, but implementation details vary by technology and jurisdiction.
Patient-generated data also introduces compliance challenges that differ from those at a traditional clinic. A research coordinator at a hospital can be trained to follow documentation protocols. A patient at home, managing their own diary entries between visits, may forget to complete entries on time, may enter data retroactively, or may accidentally modify previous entries. Systems that timestamp entries and lock them after submission help enforce data integrity, but they cannot fully replicate the controlled environment of a research site. As clinical trials increasingly move outside the clinic walls, the definition of what counts as a source document and how it should be managed is evolving in real time.