From a local dish to a product-specific model

A study published on 19 August 2026 examines how pathogen contamination and subsequent handling could affect the risk associated with Shanghai-style poached chicken. Researchers combined a local retail survey with laboratory growth experiments and a simulation covering dine-in and takeaway scenarios. The work reports modeled comparisons, not an observed outbreak.

The underlying survey took place in July 2025. It covered 84 samples from 13 Shanghai districts, including cooked-food counters and restaurants. Seven samples contained at least one of the three organisms being investigated: Salmonella was found in four, Staphylococcus aureus in two and Listeria monocytogenes in one. Those figures correspond to 4.76%, 2.38% and 1.19% of the sample set.

Why Salmonella became the focus

The team used a preliminary risk-ranking tool to compare the three organisms, then selected Salmonella for the detailed quantitative assessment. A strain recovered from the chicken samples was used to measure growth in the food matrix at six temperatures between 15°C and 40°C. The resulting growth model was combined with estimates of contamination, storage conditions and consumption.

Not every input came directly from measurements of this particular dish. Consumption was represented by broader ready-to-eat meat data, while transport and storage assumptions drew on published studies, observations, interviews, weather information and expert knowledge. Those choices made the model possible, but also define its limits.

Three complete handling scenarios

The simulation compared eating at the retail setting, takeaway followed by short ambient storage, and takeaway followed by longer refrigerated storage. Under the chosen assumptions, the ambient-storage scenario produced the highest estimated infection risk, followed by the longer refrigeration scenario and then dine-in consumption. The authors describe the two takeaway estimates as approximately four and two times the dine-in estimate, respectively.

This is not evidence that refrigeration itself increases risk. The scenarios differ in transport, elapsed time and cooling conditions as well as storage temperature. The refrigerated scenario included an initial cooling period, so it cannot be read as a controlled comparison between otherwise identical food kept cold and food kept warm.

The largest influence—and the weakest measurements

Sensitivity analysis identified the initial Salmonella contamination level as the strongest influence on estimated risk in all three scenarios. Transport conditions mattered particularly in the ambient-storage scenario; the home-storage temperature had greater influence in the longer refrigeration scenario.

Yet the initial contamination distribution was based on very little quantitative information. Of the four Salmonella-positive samples, only one had a directly quantifiable concentration. Three tested positive after enrichment but were below the direct-counting method’s quantification limit. Assigning model values to those samples necessarily introduced uncertainty.

A comparison for investigation, not a precise case forecast

The authors also acknowledge that the growth experiments did not cover refrigeration temperatures: predictions below 15°C required extrapolation. The model lacked an independent validation dataset, used a single sampling period and did not separately quantify all parameter uncertainty. Its estimated probabilities should therefore be treated as comparative indicators under stated assumptions, rather than precise forecasts of disease incidence.

The study’s clearest operational question concerns preventing contamination before the product reaches the consumer, including handling around the heating step and at retail. It does not establish a universally safe holding time or validate a new cooking process. Broader seasonal sampling, better concentration data and tests within the relevant cold-storage range would be needed to strengthen those conclusions.

SOURCES & TRANSPARENCYHow we work ↗