---
title: "Taylor Farms Rewrote Its Outbreak Statement Five Times in 16 Days"
description: "Why 2,000 negative cyclospora tests cannot clear a recalled lettuce product, and what Taylor Farms has left out of five rewritten statements."
dek: "A salad producer recalled bagged lettuce with no positive lab test against it, then defended itself with two thousand negative ones. The maths of that defence does not work — and the company's shifting public account keeps circling away from the question that does."
published_at: "2026-08-05T04:58:39.440Z"
updated_at: "2026-08-05T04:58:39.440Z"
tags:
  - "Consumer"
  - "Science"
  - "Policy"
source_url: "https://www.marlerblog.com/case-news/taylor-farms-has-rewritten-its-cyclospora-statement-four-times-in-sixteen-days-it-still-has-not-said-what-changed-at-that-plant-after-2013-or-why-two-thousand-negative-tests-should-mean-an/"
source_domain: "marlerblog.com"
canonical: "https://hex37.com/taylor-farms-rewrote-its-outbreak-statement-five-times-in-16-days"
---

On 17 July, one of North America's largest suppliers of bagged salad pulled shredded iceberg lettuce from the market. At that moment, no laboratory anywhere had found a parasite in any of its product. The [recall](https://en.wikipedia.org/wiki/Product_recall) rested on two other things: the pattern of who had fallen ill and where they had eaten, and the paper trail tracing their lettuce back through the supply chain.

Sixteen days later, the company's public account of that decision had been rewritten five times. Underneath the rewrites sits a question the food industry has never really settled. When a product is implicated by statistics rather than by a positive test result, what evidence could ever clear it?

The company has an answer, and it is a number: roughly two thousand samples taken in central Mexico since May, every one of them negative. It is an intuitive defence. It is also, on inspection, close to meaningless — and the reasons why lead directly to the questions the five versions have never once addressed.

## A parasite with exactly one hiding place

[Cyclospora cayetanensis](https://en.wikipedia.org/wiki/Cyclospora) is a single-celled parasite that causes weeks of severe diarrhoea, cramping and exhaustion. Two features make it unusual among foodborne threats.

The first is its reservoir. Salmonella lives in animals; Listeria lives in factory drains. Cyclospora lives in people. Human faeces is the only place on earth it reproduces. If it turns up on lettuce, the route it took is not a mystery in principle, only in detail: sewage, a septic system, wastewater applied to land, contaminated irrigation water, or an infected worker in a field without adequate sanitation.

The second is that nothing kills it once it is on the leaf. The US Food and Drug Administration states plainly that chlorine and the other antimicrobial rinses the produce industry relies on do not work against it. Bagged salad is a ready-to-eat product: nobody cooks it, and there is no step anywhere between the field and the bowl that reliably destroys this organism.

The parasite also takes about a week to make anyone ill, and days more before a doctor thinks to test for it. By the time an outbreak becomes visible, the lettuce that caused it has been eaten and the bags thrown away. That lag is why investigators reach for epidemiology and traceback first — and why a positive laboratory hit on the actual contaminated product is a rare piece of luck rather than a normal step.

## Why two thousand negatives is not a clean bill of health

The FDA's validated method for finding cyclospora on leafy greens is a real, published laboratory procedure. Canada's food inspection agency independently checked how well it performs. Working with greens deliberately spiked with two hundred parasite eggs, the method found them 93 per cent of the time. Roughly one contaminated sample in three came back clean. A 2023 modelling study in the *Journal of Food Protection* reached the same conclusion in more general terms: these techniques miss contamination when the levels are low.

That is the crux. A test with a substantial [false-negative](https://en.wikipedia.org/wiki/False_positives_and_false_negatives) rate, applied to a small sample of an enormous harvest, run overwhelmingly after the contamination window of May had already closed, found nothing. Two thousand of those results do not multiply into proof. They describe the sensitivity of the programme, not the condition of the lettuce.

The company's water testing is weaker still, and it says so itself: water sources are checked for indicator organisms. Indicators are common gut bacteria used as a proxy for faecal contamination in general. The FDA's own guidance for growers states that this kind of testing will not reveal the presence of cyclospora. It can flag poor water quality, which may in turn point to human waste, and that is not nothing. But a proxy for faecal loading is not a test for a parasite whose only source is human faeces — particularly when a validated method for finding it in agricultural water has existed since 2020.

So the two negatives-based arguments collapse for different reasons. The water programme was never aimed at this organism. The product programme was, but cannot bear the weight now placed on it. Which means that if the company wants to mount a defence, it has to come from somewhere else — from the regulatory record showing what it does to prevent the hazard, not from tests showing it failed to find one.

## The instruction issued in November 2013

That record exists, and it starts thirteen years ago.

In the summer of 2013, cyclosporiasis sickened 631 people across 25 states. Health authorities in Iowa and Nebraska, working with the [FDA](https://en.wikipedia.org/wiki/Food_and_Drug_Administration) and the CDC, tied 239 of those illnesses to bagged salad mix from the same company's plant at Doctor Mora, in the Mexican state of Guanajuato, served at Olive Garden and Red Lobster restaurants. Investigators later collected some 835 product, water and environmental samples there, including more than 269 human stool samples from sanitary facilities on the associated ranches. They recovered the parasite from none of them — having started work five weeks after the last known illness.

The assessment nevertheless produced an instruction. The firm was told to determine whether cyclospora is a "reasonably likely" food safety hazard in that growing region, and if so, to re-examine its washing step. That phrase is not casual. Under the US preventive-controls regime, a hazard classified that way must be met with a documented control, plus monitoring, verification and records proving the control works. The company's own materials state that its plants operate under those rules. The FDA, meanwhile, lists Mexico among the countries where cyclosporiasis is endemic, and has for years.

There is one more item from 2013. The company halted shipments to the US that August and resumed them a fortnight later with the FDA's agreement — a decision the agency said rested partly on its review of a cyclospora product sampling plan the firm had put in place. A sampling plan for this specific parasite, examined by a federal regulator, as a condition of reopening. No results from it appear to have ever been published.

So the question is not why the company failed to test. It is what its hazard analysis concludes about cyclospora, and what documentation sits behind that conclusion. If the parasite is a reasonably likely hazard at Doctor Mora, the preventive control and its verification records should exist and could be produced. If the company decided it is not a reasonably likely hazard — at the same plant, in a country its regulator calls endemic, growing a product with no kill step — then the reasoning behind that decision is the single most important document in this outbreak.

Neither has been offered.

## What moved, and what did not

Against that background, the sequence of rewrites reads differently.

The first version, on 17 July, opened with the people who had become ill and stated that product was being removed on the basis of information the FDA had supplied the day before. It also identified a specific independent farm, described as under one per cent of the American iceberg supply, as the traceback's potential source. The next day the FDA reported a positive test; the day after that it withdrew the result as a false positive.

The 19 July version led with the claim that the agency had apologised. Sympathy for the sick dropped to the fourth paragraph. The recall became an act of caution, in the past tense. The named farm disappeared. The FDA's public correction contains no apology — and in the same passage, the agency says it is continuing to work with the firm to ensure implicated product has been removed. Days later it added that the false positive changed nothing about the epidemiological basis for the recall, and that the data pointed to shredded iceberg from the company's central Mexico locations. Plural.

By 24 July the page led with money: more than $200 million a year in independently audited food safety programmes. The apology claim was gone, with no correction noting it had been there. On 30 July the material moved to a dedicated information hub; a revision on 31 July left the page contradicting itself, listing 28 recipient states in one section and 27 in another, with West Virginia — one of the nine states in the federal outbreak — present in the first and absent from the second.

The same hub presents the company's wash system, verified by the US Department of Agriculture, as a credential. It is a wash step validated against bacteria, offered in a cyclospora outbreak, by a facility that was told to re-examine that step for this parasite in 2013.

## A standard of proof that runs both ways

The company's stated position is that epidemiology can suggest a possible source but only a positive laboratory result can prove a specific product made someone sick. That is a coherent view of evidence; plenty of scientists hold it.

But it is not available selectively. A standard strict enough to discount converging traceback and outbreak data is also strict enough to discount two thousand negatives from a method that misses roughly a third of contaminated samples. If only positives count, the negatives count for nothing, and the defence dissolves into the same evidentiary void it invokes.

What is left is the gap. The company has volunteered a number that cannot do the work asked of it, and has stayed silent on the records that could. Whether that silence conceals an absence of documentation or merely a decision not to share it is not something an outsider can determine. What is observable is that five versions of the same statement have moved considerable ground — dropping a named farm, adding and then quietly deleting an apology the record does not support — without once touching the hazard analysis, the preventive control, or the sampling plan a federal agency accepted thirteen years ago as the price of reopening the plant.

*The sequence of statement revisions was documented by a plaintiffs' lawyer who has brought foodborne-illness cases since 1993, [in a detailed timeline published on his blog](https://www.marlerblog.com/case-news/taylor-farms-has-rewritten-its-cyclospora-statement-four-times-in-sixteen-days-it-still-has-not-said-what-changed-at-that-plant-after-2013-or-why-two-thousand-negative-tests-should-mean-an/).*