The chronology problem: why every IME report starts with hours of file assembly.

The opinion in a medico-legal report might take an hour to set down. The evening goes on everything before it: several hundred pages of records read end to end, events dated and cross-checked, and the one page where the MRI report actually said what you remember it saying, found again so it can be cited. That assembly work is invisible in the finished document, and yet it decides the report's accuracy, its turnaround, and how well it survives scrutiny. It is also the part of the work that has nothing to do with clinical judgement, which is exactly why it is worth examining as an operational problem rather than accepting as the price of the job.

The file, not the opinion, is the workload

A brief rarely arrives as a tidy, ordered bundle. It arrives as PDFs of scanned faxes, photographs of paper notes, hospital printouts, and sub-briefs from different parties that overlap without agreeing, and somewhere inside it sit the clinical facts the report will stand on. Before any of the doctor's actual expertise is engaged, someone has to read all of it, work out what each document is, decide which of three copies of the same consultation note is the operative one, and turn a pile of pages into a picture of what happened and when.

So the economics are lopsided by design: the instructing party is paying for the opinion, but the hours are consumed by the archaeology underneath it. That archaeology cannot be skipped, because an opinion built on a misread file is wrong no matter how good the clinical reasoning is.

The chronology is the report's evidentiary spine

It is tempting to file chronology-building under administration, but the expert evidence rules say otherwise. Under the Federal Court's expert evidence practice note and the Harmonised Expert Witness Code of Conduct, a report must set out the facts and assumptions on which each opinion is based and the reasoning that leads from them to the conclusion.¹ New South Wales binds experts to the same discipline through the code of conduct in Schedule 7 of the Uniform Civil Procedure Rules, and a report may not even be received in evidence unless the expert acknowledges it.²

Those facts are dated events extracted from the records, and the chronology is where each one acquires its date and its source. Every later sentence in the report stands on that spine, which means an error made quietly at the assembly stage, a wrong date, a conflated entry, a document misattributed, surfaces loudly later, in the least forgiving setting there is.

The same logic gives the documents-reviewed table its quiet importance. Listing the materials as D1 to Dn with dates and page counts looks like formality, but it is the report's index: it defines the universe of facts the opinion is entitled to draw on, and it is the first thing a careful reader reconciles against the letter of instruction.

A table that misses a document that was supplied, or lists one that never appears in the reasoning, invites exactly the line of questioning a report writer least wants, because it suggests the file and the opinion passed each other without properly meeting. Built well, the table and the chronology are two views of the same assembly: one lists what was read, the other shows what the reading found, dated and sourced.

Why it eats an evening

The hours go where the records disagree. The claim form says the injury was reported the day it happened; the employer's incident register says eleven days later; the first GP note mentions symptoms "for some weeks". None of those can simply be transcribed, because the report has to present each dated assertion with its source and let the discrepancy stand visibly rather than smooth it over.

Then there is the work the file makes for itself. Undated faxes have to be placed by internal evidence. The same consultation appears in the GP's notes, the insurer's summary and a specialist's letter, each version slightly different. And the work compounds: every supplementary question months later reopens the file, because "where did that date come from?" can only be answered by finding the page again, and human memory of a 400-page brief has a short half-life. A practice that writes across schemes carries this cost on every single matter, and it scales with file size, not with clinical complexity.

The hardest part of the assembly is noticing what is not there, because absence never announces itself. A chronology with a fourteen-month silence in the middle of a treatment course is telling you something, but it cannot tell you what: either the person stopped treating, or the brief is missing a tranche of records, and the report has to say which of those it cannot know.

The same is true of the smaller absences: radiology reports that reference films never supplied, a specialist letter that replies to a referral not in the file, certificates that skip a period. These gaps only become visible once every document is placed on the one timeline, which is precisely why a chronology assembled casually produces reports that quietly assume a completeness the file never had.

Page references turn verification into a lookup

The structural fix is old-fashioned scholarship: every chronology entry carries the document and page it came from, so the entry reads "12 Aug 2024 - MRI lumbar spine reported, disc protrusion L4/5 (D4, p 2)" rather than a bare assertion. The gain sits in everything that happens after the writing.

• Verification stops being a re-read and becomes a lookup. A reviewer, or the doctor at signing, moves from a sentence to its source in seconds.
• Quality assurance becomes possible at all, because a claim carrying no reference is now visible as exactly that.
• The honest gaps stay honest. "Report sighted, images not reviewed" is easy to preserve when the reference trail shows precisely what was and was not in the file.
• Supplementary rounds stop being archaeology, because the trail built the first time answers the follow-up question.

So the same discipline that makes a report defensible under cross-examination, every fact traceable to its page, turns out to be the discipline that makes the practice efficient.

There is a staffing consequence hiding in this, and for a medico-legal network it is the significant one. An unreferenced draft can only be checked by the person who read the whole file, which in practice means the doctor checks their own assembly, at the end of the day, against their own memory, which puts the most expensive reviewer in the building on the least clinical part of the work, in the conditions most likely to produce a miss.

A referenced draft can be checked by anyone. A practice manager can verify that every date matches its cited page without any clinical knowledge at all, and the doctor's review narrows to the judgements that actually need a doctor. Page references do not just speed the checking up; they change who is able to do it.

The rules expressly bless tooling for this layer

If the assembly layer is where the hours go, the natural question is whether software is allowed to carry it, and for once the regulatory answer is unambiguous. The NSW Supreme Court's generative AI practice note, the strictest instrument in the country, expressly permits using generative AI for chronologies, indexes and the summarising of documents, even while requiring leave before AI drafts expert report content.³ The Personal Injury Commission's PIC13 goes the same way, expressly permitting the generation of chronologies, indexes and schedules and the summarising or review of documents and transcripts.⁴

Even the cautious professional-indemnity guidance points in this direction: Avant's advice to report writers is to list the information received and answer the questions asked with reasons, which is precisely what a rigorous, source-referenced assembly makes easy.⁵ The regulated zone is generative drafting of report content, and the forbidden zone is the opinion itself; the assembly layer sits in the expressly permitted zone in every Australian forum that has spoken.

Treat the chronology as infrastructure

Treat the chronology as infrastructure and hold it to infrastructure standards: every entry dated and referenced to its document and page, inconsistencies surfaced side by side rather than silently resolved, missing records named as missing, and the whole assembly reproducible months later when the supplementary questions arrive.

That is the design brief Trenthos IME is built against. It ingests the brief as it arrived, scans included, runs OCR onshore, and builds a dated chronology with a page reference on every entry, then drafts the report with each factual line linked to its source. The doctor's evening goes to the examination and the opinion, which are the parts only a doctor can supply and the parts the software is designed never to touch.

The chronology problem does not need heroics to solve. It needs the ordinary work done systematically, once, with references, so that everything downstream, the review, the signing, the cross-examination, the follow-up, draws on the trail instead of the doctor's recall of page 214.