Discordant cytology audit: disclosure before reassurance

Frame it first. This is a patient safety incident with potential serious harm, not a documentation issue. That framing decides who is told, how fast, and at what level.

1. Stop the draft letter. 'Minor discrepancies' prejudges severity. Nothing leaves the building that characterises the finding before independent verification, because a later contradiction destroys credibility with patients and staff.
2. Clinical first, system second. Establish the current status of the four women: are they in colposcopy follow-up, treated, discharged, alive? For the deceased woman, identify next of kin and whether anyone has already spoken to them. A woman with an untreated high-grade lesion needs clinical review this week, not next month.
3. Independent verification. Ask a second laboratory or external expert to re-review the nine slides blind. Nothing is communicated as established fact until that is complete, but the gap between internal report and verification is measured in days, and you record why any longer.
4. Governance. Convene an incident management team (clinical lead, laboratory, quality, communications, legal, data protection). Report through the HSE incident management system, notify the National Screening Service quality function, and agree one point of contact.
5. Communications and records. Holding statement only; no confirmation of numbers or causes to media. Keep a contemporaneous decision log.

What I would not do: suspend the whole programme. Withdrawing screening from thousands of women carries more aggregate harm than the laboratory problem being investigated. Contain the specific process instead.

Disclosure is a process, not a letter, and it follows from verified facts. For each of the four:

• Who and how. A senior clinician plus a second person, in person or by phone if travel is difficult, with a written summary afterwards. Never a standardised letter alone for a potentially significant under-call.
• Content. A plain statement of what the audit found, what it means for her (an earlier result may have missed a change that was present), an apology on behalf of the service where the finding is confirmed, what is happening now (clinical review, colposcopy, treatment if needed), and what will happen next with named contacts.
• No speculation about blame or outcome. Do not say 'nothing would have changed' or 'this caused the death'; those conclusions belong to the verified review and, where relevant, to a separate process.
• Offer practical and psychological support - a named liaison, counsellor access, help with travel.
• The deceased woman's family: contact next of kin sensitively, offer a meeting with a senior clinician, explain the audit finding and what it does and does not tell them about her death, and avoid implying causation before the review concludes. Be ready for questions about relatives - cervical cancer is not inherited, but her daughters may be eligible for screening, and that can be discussed factually.
• Record what was said, by whom, when, and any follow-up promised. Check each woman knows the audit involved her records; if consent for audit use was not clearly covered in programme information, address that.

Consistency matters more than perfect wording: all four get the same core explanation, adjusted to their clinical situation.

The failure here is not only the missed slides - it is that an out-of-threshold interval cancer rate apparently did not trigger escalation until an audit was completed. So the design must detect drift earlier.

Indicators:

• Interval cancer rate against the 1.4 per 1,000 threshold, reported monthly with a defined escalation trigger (for example, two consecutive months above threshold, or any single month above a higher bound).
• Blind re-review discordance rate with graded action thresholds - minor discordance for education, significant under-call for mandatory incident review.
• Correlation between cytology and histology for all high-grade outcomes.
• Proportion of negative reports in women later diagnosed with CIN3 or worse, by grader and by year.
• Grader certification, continuing education and annual competency review.

Assurance: an annual external audit by a laboratory not involved in the service; a KPI dashboard that goes to the programme board, not only the laboratory manager; a written escalation standard that names who is notified within 24 hours of a breach; and a just-culture route for staff to raise concerns about reporting pressure without fear.

How I would know it worked: the interval cancer KPI returns to and stays within threshold across two consecutive audit cycles; discordance falls and remains inside the agreed band; the next KPI breach is escalated within the standard without waiting for an audit; and staff can describe the escalation route and say they have used it. I would also test the system deliberately - a tabletop exercise with a simulated breach - because a route that exists on paper is not evidence it works.