A laboratory is asked to amend a report issued eighteen months earlier, after a variant in it is reclassified. The scientist opens the original: a document written in a word processor from a template, with the variant table pasted in from the pipeline and the interpretation written freehand. Which candidates were considered and rejected is not in it. Which database releases were used is not in it. Why the second-ranked gene was dismissed is in an email, perhaps.
The amended report has to be written from the old report, because the old report is the only record there is.
What the field knows
The reporting standards describe what a report must contain and say little about where it comes from. The ACMG laboratory standards for next-generation sequencing (Rehm et al., 2013) require the report to state the methods, the regions analysed, the variants found with their classification and the evidence, and the limitations of the test. The ACMG and AMP variant interpretation framework (Richards et al., 2015) defines the classification and the evidence categories behind it. European recommendations for reporting genomic results (Deans et al., 2022) ask for the same elements and add clarity on which findings were and were not sought.
What the standards share is an assumption that the content is available to be stated: the evidence was gathered, the alternatives considered, the versions known. Vears and colleagues (2017), reviewing reporting practice for variants of uncertain significance, found wide variation in what laboratories put in a report and how, with the same evidence base producing documents of very different completeness. The content existed somewhere; the report was a selective rendering of it by whoever wrote it, and the selection was not recorded either.
The reanalysis guidance (Deignan et al., 2019) makes the dependency explicit. A re-evaluation needs to know what the first evaluation did, and a report that records the conclusion without the reasoning cannot support one. The points to consider ask laboratories to keep the data and the interpretation in a form that can be revisited.
Outside genetics the lesson is older. A document that is both the working record and the published output ends up serving the output, because that is what people see. Systems that keep a structured record and derive documents from it can regenerate the document when the record changes and can show the difference between two documents in terms of what changed in the record.
Where practice breaks
The report becomes the source of truth by default. The pipeline output is in a file, the evidence is in browser tabs, the discussion is in a meeting, and the only durable artefact is the document that leaves the laboratory. Everything that did not make it into the document is lost when the scientist's memory fades or the scientist leaves.
Freehand interpretation drifts from the evidence. A sentence is reworded for a clinician's benefit and the hedge in it is lost. A phrase from the last report is reused because it read well. A variant table is pasted from one pipeline version and the methods paragraph describes another. None of this is dishonest and all of it makes the report an unreliable witness to the case.
Templates age in the same way. A methods paragraph written for one capture kit and one pipeline version is reused for years after both have changed, because nobody owns the sentence and the report is assembled by copying the last one. The document describes a test the laboratory no longer runs.
Amendments compound it. Each amended report is written from the previous one, with a sentence added at the top. After two amendments nobody can reconstruct what the laboratory believed at any given date, which matters when a family asks why they were told one thing in one year and another later.
The rejected hypotheses vanish. A report states what was found. It rarely states what was considered and set aside, and why, so the next analyst re-examines the same dismissed candidates and the family is never told that an alternative was looked at. For an unresolved case, the rejected hypotheses are most of the work that was done.
On the receiving side, the report is read once and summarised again. The clinician copies its conclusion into the patient's notes as a line, the line is copied into the next referral, and the hedge that the scientist wrote carefully is lost a second time. A report whose structure is visible, with the classification, the evidence and the open questions as separate elements, survives that copying better than a paragraph does.
How Tulparus handles it
The case record is the source of truth and the report is a view of it. Every element a report contains, the samples and methods, the knowledge releases, the candidates with their evidence and classification, the hypotheses with their standing and the reasons for it, the open questions and proposed next steps, lives as structured state on the case, versioned and dated. The report is rendered from that state when a person asks for it.
The person still writes. The interpretation paragraph is authored by the scientist, with the evidence it refers to linked to the record, and every claim in it is checked against the record before the report can be signed: a classification in the prose must match the classification on the case, a version in the methods must match the version the analysis used. A report cannot say a thing the case does not know.
Sign-off is a recorded decision on a specific case version. An amendment is a new case version and a new rendered report, and the two can be shown side by side with the differences in the record that caused the differences in the document. The earlier report is kept as it was issued, and a reader of the amendment can see what the laboratory believed on each date.
Rejected hypotheses stay in the record and can be rendered into the report when the laboratory's template calls for it, as a short list of what was considered and why it was set aside. For the laboratory in the opening case, amending the report would mean opening the case, recording the reclassification against its new release, letting the hypothesis update, and rendering again. The reasoning would be where the scientist left it.
What you can do tomorrow
The habit can start before the tooling, with what is kept beside the report.
- Keep a structured case note alongside every report: candidates considered, evidence for and against each, the versions used, and who decided what. A spreadsheet is enough to begin.
- Add a line to the report template for the knowledge releases used, and fill it from the pipeline rather than from memory.
- List the rejected hypotheses in the case note, with the reason. It is the part reanalysis most needs and the part most often lost.
- When amending, write from the case note rather than from the previous report, and keep every issued version unchanged.
Sources
- Rehm HL, Bale SJ, Bayrak-Toydemir P, et al. ACMG clinical laboratory standards for next-generation sequencing. Genetics in Medicine, 2013.
- Richards S, Aziz N, Bale S, et al. Standards and guidelines for the interpretation of sequence variants: a joint consensus recommendation of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology. Genetics in Medicine, 2015.
- Deans ZC, Ahn JW, Carreira IM, et al. Recommendations for reporting results of diagnostic genomic testing. European Journal of Human Genetics, 2022.
- Vears DF, Sénécal K, Borry P. Reporting practices for variants of uncertain significance from next generation sequencing technologies. European Journal of Medical Genetics, 2017.
- Deignan JL, Chung WK, Kearney HM, Monaghan KG, Rehder CW, Chao EC. Points to consider in the reevaluation and reanalysis of genomic test results: a statement of the American College of Medical Genetics and Genomics (ACMG). Genetics in Medicine, 2019.