When the referral file contradicts itself.
A referral file is not one witness. It is many, layered across years: a GP's contemporaneous note, a specialist's letter written for a different purpose, an insurer's file summary, the claimant's own account, and sometimes a prior assessment that reached a different view. Ask each of those sources the same simple question, when did the symptoms start, and you will not always get the same answer. Long before you form your own opinion, the file has usually already disagreed with itself.
That is not a defect in the file. It is what a real medical history looks like when it is assembled from multiple hands over multiple years, none of whom were writing for you. The question is not whether the file is internally consistent. It usually is not. The question is what you do with the disagreement, and whether your report shows your working.
Every file is really several files
The pattern repeats across case types. A mechanism of injury described one way in the incident report and another way in the first medical note. A symptom that appears in a GP's file six months before the claimant says it started. Medication lists that do not match between the pharmacy record and the specialist letter. A prior assessment that found no ongoing impairment, sitting in the same bundle as a treating report that finds a substantial one. None of this means anyone lied. Memory is imperfect, notes are taken quickly, and different clinicians ask different questions for different reasons. But it does mean the file you are handed is not a single reliable narrator. It is several partial ones, and part of the job is noticing where they part company.
This is where a big file becomes more than a reading-time problem. A 300-page bundle read quickly is not just at risk of missed pages. It is at risk of missed disagreements, because a contradiction that sits fifty pages apart is much easier to miss than one on the same page.
This is not just an impression from doing the work. A 2024 systematic review in the Journal of Health Management screened more than 7,600 studies on medical record documentation and found inconsistency, alongside incompleteness and inaccuracy, a recurring, well-documented category of error in the literature. The referral file disagreeing with itself is not a sign that something unusual has gone wrong with your particular case. It is closer to the default state of a record built by many hands.
The code does not let you pick the tidy version
Under the Expert Witness Code of Conduct in Schedule 7 of the Uniform Civil Procedure Rules 2005 (NSW), and the equivalent harmonised code annexed to the Federal Court's expert evidence practice note, an expert's report must state the facts and assumptions on which the opinion is based, and where an opinion is not fully researched, the expert has to say so. Read literally, that duty is not just about disclosing your own reasoning. It is about disclosing the state of the record you reasoned from. An opinion that quietly adopts one version of a contested history, without noting that the file contained another, has not fully stated the facts and assumptions it rests on. It has stated the convenient ones.
The same principle sits behind the common law test for whether an opinion is admissible at all. In Makita (Australia) Pty Ltd v Sprowles [2001] NSWCA 305; 52 NSWLR 705, the leading Australian authority on expert evidence, the NSW Court of Appeal held that an expert must make explicit the facts observed or assumed and the reasoning that connects them to the conclusion, so the court can independently test it rather than take the opinion on faith. An opinion whose factual basis is left unclear is not just weaker for it. Left unexplained, it risks being ruled inadmissible, or given little weight even where it is let in.
That is a different failure from getting the clinical judgement wrong. You can be entirely correct in your opinion and still leave a report exposed, because the exposure is procedural, not clinical: the reader cannot see that you knew about the conflicting version and can no longer test whether you dealt with it.
A report that never mentions the conflicting version does not look more certain. It looks like nobody checked.
Why the instinct is to smooth it over
Under time pressure, the fastest path through a contradictory file is to pick the version that best supports a clean narrative and keep moving. It is an understandable instinct. Naming every inconsistency feels like it slows the report down and invites questions you would rather not prompt. But that instinct has the risk backwards. A skilled cross-examination is built on the assumption that the expert has read the whole file. If the barrister on the other side has found the contradiction and you have not mentioned it, the conversation stops being about the clinical question and becomes about what else you might have missed. The report's credibility, not just the disputed fact, is what ends up on trial.
This is not a hypothetical risk in Australian medico-legal practice. In Kemp v Insurance Australia Limited t/as NRMA Insurance [2024] NSWPICMP 260, a NSW Personal Injury Commission Review Panel weighed a psychiatric opinion against the contemporaneous clinical records and found the history behind it did not hold up against them, an illustration of a principle Medical Assessors apply routinely: an opinion resting on a history that conflicts with the contemporaneous record tends to carry less weight than the record itself. A finding that is inconsistent with the documented history, without the reasoning to explain why, is itself recognised ground for a Commission review. The disagreement in the file does not stay buried. It resurfaces at exactly the moment it can do the most damage to the opinion that ignored it.
The opposite approach, naming the conflict and explaining how you resolved it, costs you a paragraph. It also removes the opening. There is no discrepancy left for opposing counsel, or a review panel, to produce as a surprise, because you produced it first, on your own terms, with your own reasoning attached.
What noting it actually buys you
An opinion that says the file is unclear on a specific point, and explains how that uncertainty was weighed, is not a weaker opinion than one that states a fact with false confidence. It is a more defensible one, because it demonstrates exactly the reasoning process the expert evidence rules ask for: the facts considered, the assumptions made where the facts conflicted, and the limits of what the record actually supports. Certainty you do not have is not a strength you can bank. It is a liability waiting for someone else to find it first.
- Name the conflict. State plainly that two sources in the file disagree, and what each one says.
- Show the resolution. Explain which account you relied on and why, whether that is contemporaneity, clinical plausibility, or corroboration elsewhere in the file.
- Flag what is still open. If the conflict genuinely cannot be resolved from the file alone, say so rather than silently choosing a side.
How Medilee is built for this
This is one of the reasons Medilee's case summary does not just extract a chronology, it automatically flags inconsistencies and discrepancies as it builds one, so a conflicting date, a symptom onset that does not match across sources, or a mismatched medication list surfaces during your review instead of surviving unnoticed into the draft. Every flagged item is source-linked back to the exact pages it came from, so you are not taking Medilee's word that a conflict exists. You are confirming it against the record, in seconds, the same way you would confirm any other fact. The judgement about what the conflict means, and how it affects your opinion, stays entirely yours. What changes is that you get the chance to make that judgement deliberately, on every file, rather than only on the contradictions you happened to catch on a fast read.
A note on scope
The expert evidence obligations, and the cases and studies referenced here, illustrate a principle rather than guarantee an outcome in any particular matter. The exact rules differ by court, tribunal and scheme, and case law develops. This is not legal advice, and the underlying principle is what matters regardless of the specific rule: an opinion is only as strong as the record it is honest about, and a file that disagrees with itself is not a reason to hide the disagreement. It is a reason to show it.
Sources
- Shahbodaghi, A., Moghaddasi, H., Asadi, F. and Hosseini, A. (2024). Documentation Errors and Deficiencies in Medical Records: A Systematic Review. Journal of Health Management (SAGE).
- Expert Witness Code of Conduct, Schedule 7 of the Uniform Civil Procedure Rules 2005 (NSW).
- Makita (Australia) Pty Ltd v Sprowles [2001] NSWCA 305; 52 NSWLR 705, as discussed in the NSW Judicial Commission's Civil Trials Bench Book, "The opinion rule".
- Kemp v Insurance Australia Limited t/as NRMA Insurance [2024] NSWPICMP 260 (29 April 2024), case summary: Carroll & O'Dea Lawyers.
- Bartier Perry, "Assessing new injury claims in existing NSW workers compensation matters: the importance of identifying inconsistencies and the role of contemporaneous evidence", on the Personal Injury Commission's review grounds for findings inconsistent with the documented history.
Case summaries are cited from the legal commentary linked above rather than the primary judgment. Check the current version of any source before relying on it for your own matter.
