The Chicago Department of Public Health (CDPH) conducts routine food inspections of over 15,000 food establishments to ensure the health and safety of their patrons.
In 2015, CDPH deployed a machine learning model to schedule inspections of establishments based on their likelihood to commit critical food code violations.
The City of Chicago released the training data and source code for the model, allowing anyone to examine the model. We provide the first independent analysis of the model, the data, the predictor variables, the performance metrics, and the underlying assumptions. We present a summary of our findings, share lessons learned, and make recommendations to address some of the issues our analysis unearthed.
Hindsight analysis of the Chicago food inspection forecasting model, 2019
Illinois Institute of Technology
Vinesh Kannan, Matthew Shapiro, and Mustafa Bilgic
Danielle Ann of Alvinology reports researchers at the Singapore General Hospital have found definite similarities between the virus strains of Hepatitis E virus or (HEV) in pig liver and human liver.
This means that ingesting raw pork liver could mean you’re ingesting a strain of HEV that’s similar enough to human HEV that it could cause you get infected.
The same report said that people who have contracted HEV has risen steadily over the years. While the researchers could not say if the ingestion of raw pig liver is the main cause of the rise in cases, many local dishes feature this ingredient and do not cook the meat thoroughly.
The same report said that you can acquire the disease from eating contaminated food or substances. Ingesting water that is laced with the disease or accidentally drinking water that has trace amounts of faeces. Eating raw or half-cooked meat that is infected can also transmit the virus to you.
Rebecca Trager of Chemistry World reports U.S. researchers have created a handheld detection system that is sensitive enough to catch just a few particles of norovirus.
University of Arizona biomedical engineer Jeong-Yeol Yoon and his team have created a highly sensitive portable detection system capable of spotting norovirus at levels that can make people sick. The work was presented the American Chemical Society’s national meeting in San Diego, California on 27 August.
As few as 10 norovirus particles can cause vomiting and diarrhoea in humans and the virus is extremely contagious so early detection is vital to prevent outbreaks. However, the virus does not grow in laboratory cultures and current detection methods rely on specialised and time-consuming PCR (polymerase chain reaction) techniques.
Yoon’s research team previously developed a smartphone-based device that measured light-scattering from norovirus-bound polystyrene beads in a paper microfluidic chip. It has now improved the device’s detection limit by changing to a fluorescence-based method.
‘I looked at Amazon.com and saw that they sell a lot of these smartphone attachments – smartphone microscope attachments – that turn your phone turns into a microscope, and by adding a couple of other components, I could convert the smartphone-based microscope into a fluorescence microscope,’ Yoon explains.
The setup uses a microscope accessory with a separate light source and two optical filters. He and colleagues also designed a 3D printed case to house the components.
To test a sample, it is first added to the paper microfluidic chip, followed by a suspension of fluorescent beads labelled with norovirus antibodies. After three to five minutes, the antibodies bind to any norovirus particles in the sample, creating aggregates of the fluorescent beads that spread out along the channels of the chip. The resulting increase in fluorescence intensity around each norovirus particle can be detected by taking a picture of the chip with the smartphone’s camera.
An app that the team has also developed then analyses the picture to calculate the sample’s norovirus concentration from the pixel count in the image. So far, the lowest detection limit corresponded to about 5 or 6 norovirus particles per sample, Yoon says. He estimates that the material costs of this system, aside from the cell phone and app development costs, are about $200.
I love my shellfish – mussels and scallops (oysters are sorta gross) – but the kid was diagnosed with a shellfish allergy so they’re an occasional lunch while she’s at school.
Health types in the UK report on six cases of diarrhetic shellfish poisoning (DSP) following consumption of mussels in the United Kingdom (UK). The mussels contained high levels of heat-stable okadaic acid (OA)-group toxins. Here we describe the environmental and epidemiological investigation carried out in response to the outbreak.
In June 2019 (day 0), Public Health England South West was notified by the local authority of three diners who were unwell following consumption of mussels in a restaurant 5 days earlier. The local authority had determined that the restaurant had had received a batch recall notice, also 5 days earlier, from the shellfish producer for the mussels because of elevated toxin levels but this was not seen before the mussels were served that day. On day 1, PHE South West received a report from the county neighbouring the first of gastrointestinal illness linked to mussels from the same producer. A multi-agency outbreak control team was therefore convened on day 2 and led by the PHE South West health protection team.
An alert was sent to all health protection teams across England on day 2 asking about any reported cases of gastrointestinal illness following consumption of mussels. Local authorities in areas of product distribution were informed of the identified risk by email. Persons reporting illness who were identified by local authorities as having consumed the affected mussels were asked by PHE to complete a bespoke questionnaire on exposure and clinical data.
A probable case of DSP was defined as an individual with diarrhoea, three or more loose stools in 24 h, or vomiting or abdominal cramps or nausea, with date of onset from 7 days before to 1 day after notification of the outbreak, and time of onset 30 min to 24 h following consumption of mussels harvested from the affected site. Confirmed cases were as probable, but with an absence of pathogens in a stool sample that would otherwise explain illness.
Thirteen individuals reported to have been unwell after consumption of mussels were contacted. Completed questionnaires were received from seven individuals, of which three were confirmed, and three probable cases. The cases ate at four separate venues. One respondent did not meet the case definition as symptom onset was more than 24 h following consumption.
The mean age of cases was 59 years (range: 37–76 years); three were male and three were female. All cases reported eating steamed mussels. Five cases ate mussels as a main course and one as a starter. Reported portion sizes ranged from 11 to 50 mussels.
The mussels were produced in an offshore marine area. A routine shellfish monitoring programme is in place throughout England and Wales, including at the affected site. As a part of this programme, the water column is sampled every 2 weeks from April to September and cell counts of potentially harmful algal species are measured. Shellfish flesh samples are also tested for the presence of selected European Union (EU)-regulated biotoxins every 4 weeks during April to September each year unless phytoplankton counts and/or shellfish toxins are quantified above specified warning limits that require further precautions, including re-testing and closure.
Lipophilic toxin determination, including that for OA-group toxins, is routinely carried out using the method specified in in the EU-Harmonised Standard Operating Procedure for determination of lipophilic marine biotoxins in molluscs by LC-MS/MS . Additional flesh and water samples were taken in advance of the planned sampling date following a report to the local authority from a local fisherman of a red-coloured algal bloom six miles offshore from the production site.
The local authority determined the source of the mussels by questioning venues linked to reports of illness. Subsequently, the shellfish producer provided the outbreak control team with a complete list of all businesses who had received the affected mussels. Mussels from the site were harvested daily from 9 to 5 days before notification of the outbreak for commercial sale. The mussels were not tested by the producer for the presence of toxins. A large volume of mussels was distributed to seafood wholesalers, restaurants and pubs, and subject to the recall notice distributed by the producer 5 days before reports of illness to PHE. A limited number of businesses not linked to any known cases, including wholesalers, retailers, restaurants and pubs, responded to the recall stating they had sold some of the affected produce. No produce was found to still be in circulation at the time of the outbreak response.
Water column and shellfish flesh sampling results are summarised in Table 2. Measured densities of Dinophysis spp. in the water column increased rapidly from being undetectable 16 days before outbreak notification to 1,600 cells per litre 7 days before, coinciding with the time of harvesting of the affected batch and exceeding the England, Wales and Northern Ireland Food Standards Agency trigger level of 100 cells per litre. The level of total OA-group lipophilic toxins in mussel flesh was 338 µg OA equivalents (eq) per kg, following application of measurement uncertainty, 7 days before outbreak notification. This exceeded the maximum permitted limit (MPL) of 160 µg OA eq per kg defined by European Commission (EC) regulation 853/2004 . Toxin concentrations quantified showed that an average of 94% of the OA-group toxins present in the mussels consisted of OA itself, with the remainder being dinophysistoxin 2 (DTX2).
Water column sampling 7 days before outbreak notification did not detect other harmful algal species apart from Pseudo-nitzschia spp., the causative diatom for domoic acid responsible for amnesic shellfish poisoning, at 1,320 cells per litre. This is below the trigger level of 150,000 cells per litre for this species.
Routine shellfish sampling at the same site during the same time period did not detect paralytic shellfish poisoning toxins. Trace levels of yessotoxins were detected, but along with traces of azaspiracids, they were well below regulatory levels. Amnesic shellfish poisoning toxins were below the limit of quantitation (LOQ).
In response to the elevated toxin levels quantified and reported 5 days before outbreak notification, the shellfish bed was immediately closed for harvesting as per standard practice in England. The Food Standards Agency urgently contacted local authorities of places where the affected product had been distributed to ensure that wholesalers and venues had acted upon the recall. Venues were asked whether any product had been frozen, for example in the form of stock, as this would not deactivate the toxin, but there was no evidence this had been done.
We report on six cases of DSP associated with consumption of mussels harvested in the South West of England. Without an available validated test for relevant toxins in human samples, the diagnosis was made based on characteristic clinical symptoms, including diarrhoea, abdominal pain, nausea and fever/chills, elevated levels of OA-group toxins in the flesh of mussels from the same batch as those consumed, the absence of faecal pathogens in stool of cases and epidemiological evidence of exposure to the produce.
DSP occurs following consumption of seafood containing high levels of the heat-stable OA-group toxins produced by dinoflagellates including Dinophysis spp., and is characterised by a rapid-onset of self-limiting gastrointestinal illness [3,4]. Recognised outbreaks of DSP are rare. Seventy cases were identified in 2013 following consumption of mussels harvested around the Shetland Islands  and 49 cases were identified in 1998 following consumption of UK-harvested mussels in London . Outbreaks have been recorded in recent years in China, the United States, France and Canada [4,7–9].
The lowest-observed-adverse-effect level of OA is 45 to 50 µg OA eq per person [4,10]. In our study, an average main course portion of mussels (500 g in shell) would provide 41 µg OA eq., using a flesh weight yield of 24% . This level of exposure is consistent with DSP as the cause of illness considering variability in portion sizes, flesh yield, body weight and toxin levels at the production site. Individual mussel sizes served were unavailable but would likely vary. Therefore, overall estimated portion weight was used to calculate the exposure dose. A limitation is that body weight (bw) was not recorded for cases and because of this, OA eq per kg bw could not be calculated.
A shellfish biotoxin programme monitoring the occurrence of harmful algal blooms and toxins in classified shellfish production areas in the UK, alongside food business operator checks, remains a robust system to protect population health. Nonetheless, a rapid increase in concentrations of Dinophysis spp. cells within the waters of the production site may have contributed to the outbreak, in tandem with shellfish harvesting occurring before official control results were reported and site closure. Whyte et al. (2014) demonstrated that a similar rapid increase in Dinophysis levels, resulting from a change in prevailing wind direction, occurred in the 2013 Shetland Islands origin outbreak . Transdisciplinary research is required to predict future risk and inform monitoring, particularly given likely changes in the distribution of potentially-toxic species particularly if temperature of ocean water increases . Our investigation suggested that affected produce may have been sold by restaurants and pubs with no known linked cases. Given that DSP is a self-limiting illness that may be under-reported by cases and has low awareness among clinicians, the actual number of persons affected in this outbreak is likely to be higher .
This outbreak highlights that clinicians and public health professionals should be aware of algal-derived toxins as a potential cause of illness following seafood consumption, and that the need for effective end-product testing of shellfish to ensure food safety remains.
Outbreak of diarrhetic shellfish poisoning associated with consumption of mussels, United Kingdom, May to June 2019
Due to the potential of enterohemorrhagic Escherichia coli (EHEC) serogroup O157 to cause large food borne outbreaks, national and international surveillance is necessary.
For developing an effective method of molecular surveillance, a conventional method, multilocus variable-number tandem-repeat analysis (MLVA), and whole-genome sequencing (WGS) analysis were compared. WGS of 369 isolates of EHEC O157 belonging to 7 major MLVA types and their relatives were subjected to comprehensive in silico typing, core genome single nucleotide polymorphism (cgSNP), and core genome multilocus sequence typing (cgMLST) analyses. The typing resolution was the highest in cgSNP analysis. However, determination of the sequence of the mismatch repair protein gene mutS is necessary because spontaneous deletion of the gene could lead to a hypermutator phenotype. MLVA had sufficient typing resolution for a short-term outbreak investigation and had advantages in rapidity and high throughput. cgMLST showed less typing resolution than cgSNP, but it is less time-consuming and does not require as much computer power. Therefore, cgMLST is suitable for comparisons using large data sets (e.g., international comparison using public databases). In conclusion, screening using MLVA followed by cgMLST and cgSNP analyses would provide the highest typing resolution and improve the accuracy and cost-effectiveness of EHEC O157 surveillance.
IMPORTANCE Intensive surveillance for enterohemorrhagic Escherichia coli (EHEC) serogroup O157 is important to detect outbreaks and to prevent the spread of the bacterium. Recent advances in sequencing technology made molecular surveillance using whole-genome sequence (WGS) realistic. To develop rapid, high-throughput, and cost-effective typing methods for real-time surveillance, typing resolution of WGS and a conventional typing method, multilocus variable-number tandem-repeat analysis (MLVA), was evaluated. Nation-level systematic comparison of MLVA, core genome single nucleotide polymorphism (cgSNP), and core genome multilocus sequence typing (cgMLST) indicated that a combination of WGS and MLVA is a realistic approach to improve EHEC O157 surveillance.
Effective surveillance using multilocus variable-number tandem-repeat analysis and whole-genome sequencing for enterohemorrhagic Escherichia coli O157
Applied and Environmental Microbiology
Kenichi Lee, Hidemasa Izumiya, Sunao Iyoda, Makoto Ohnishi and EHEC Working Group
The testing was carried out on around 600 samples of lamb, beef, pork, and chicken of both Icelandic and foreign origin between March and December 2018. The purpose of the testing was to determine the prevalence of pathogenic micro-organisms in products when they reach the consumer, and for this reason the samples were taken from shops.
Campylobacter and salmonella were not detected in pork or chicken samples, with the exception of a single sample of pork from Spain. MAST attributes this to improved preventative measures in slaughterhouses.
MAST points to several ways consumers can reduce the risk of infection from salmonella, campylobacter, and E. coli, including cooking meat all the way through and taking care to avoid cross-contamination. Most E. coli is found on the surface of meat, and therefore is killed by frying or grilling, but when meat is ground, the bacteria is distributed throughout. Therefore, hamburgers and other types of ground meat should be cooked through.
The collection of comprehensive data on both food exposure history and gastrointestinal symptoms associated with foodborne illness is typically challenging because of poor compliance with data collection methods. Smartphone technology provides a convenient tool with which to collect data on food consumption behavior and foodborne illness.
Ethica, a smartphone-based application used to acquire, store, and analyze data on human behavior, was evaluated as a potential supplement to current data collection strategies. The purpose of this study was to assess the compliance rates, advantages, and drawbacks of a smartphone-based method of collecting food history and foodborne illness data from a sample of volunteers.
Through a combination of user-triggered and prompted microsurveys, meal descriptions, and photo food diaries, the food consumption behavior of 96 university students was investigated over a period of 10 weeks. During the first 10 days of the study, 95% of participants used the time-triggered microsurveys and 51% of participants used the digital image features of the app to report food consumption history at least twice per day. Approximately 34% of participants used at least one of the reporting features to report at least one episode of vomiting or nausea during this period, and 29% reported at least one episode of diarrhea; only 7% sought medical care.
The smartphone-based method of data collection used in this study allowed capture of uniquely detailed food history data and data on gastrointestinal symptoms associated with foodborne illness that were not reported to medical practitioners and public health.
This enhanced ability to gather in-depth information from sentinel and at-risk target populations could support foodborne illness surveillance. Compliance rates, advantages, and limitations discussed in this study can guide the development of future data collection strategies.
Compliance rates, advantages, and drawbacks of a smartphone-based method of collecting food history and foodborne illness data, June 2019
Journal of Food Protection vol. 82 no. 6 pp. 1061-1070
Patrick Seitzinger, Nathaniel Osgood, Wanda Martin, Joanne Tataryn, and Cheryl Waldner
The roles of environmental reservoirs, including wild birds, in the molecular epidemiology of Campylobacter jejuni have not been assessed in depth.
Our results showed that game birds may pose a risk for acquiring campylobacteriosis, because they had C. jejuni genomotypes highly similar to human isolates detected previously. Therefore, hygienic measures during slaughter and meat handling warrant special attention. On the contrary, a unique phylogeny was revealed for the western jackdaw (right) isolates, and certain genomic characteristics identified among these isolates are hypothesized to affect their host specificity and virulence.
Comparative genomics within sequence types (STs), using whole-genome multilocus sequence typing (wgMLST), and phylogenomics are efficient methods to analyze the genomic relationships of C. jejuni isolates.
Population Genetics and Characterization of Campylobacter jejuni Isolates from Western Jackdaws and Game Birds in Finland Sara Kovanen, Mirko Rossi, Mari Pohja-Mykrä, Timo Nieminen, Mirja Raunio-Saarnisto, Mikaela Sauvala, Maria Fredriksson-Ahomaa, Marja-Liisa Hänninen and Rauni Kivistö
Appl. Environ. Microbiol. February 2019 85:e02365-18; Accepted manuscript posted online 14 December 2018, doi:10.1128/AEM.02365-18
Salmonella enterica is one of the principal causes of foodborne zoonotic enteritis. Among the different serovars, Dublin (S. Dublin) is of particular importance due to its propensity to progress to an invasive infection in humans and due to the high proportion of multi-drug resistant strains in Canada.
Cattle are considered as the main reservoir of S. Dublin. This serotype has emerged since 2011 in the province of Quebec, Canada, in both cattle and human populations. First animal cases have been reported in calf production.
White veal are valued for the quality of their meat, offal and liver. The liver is usually consumed mildly cooked and is considered as a probable source of foodborne exposure to S. Dublin in humans. The objective of this study was to estimate the prevalence of S. Dublin positive liver after slaughtering and the seroprevalence against S. Dublin at the calf level.
Prevalence of salmonella Dublin in veal liver in Quebec, Canada from a public health perspective, February 2019
International Journal of Infectious Diseases vol. 79 pg. 75
C.M. Andela Abessolo, P. Turgeon, P. Fravalo, G. Côté, G. Eyaba, W.P. Thériault, J. Arsenault
Official control in slaughterhouses, consisting of meat inspection and food safety inspection, has an important role in ensuring meat safety, animal health and welfare, and prevention of transmissible animal diseases. Meat inspection in the European Union (EU) includes the inspection of food chain information, live animals (ante-mortem inspection), and carcasses and offal (post-mortem inspection).
Food safety inspections are performed to verify slaughterhouses’ compliance with food safety legislation and are of the utmost importance, especially if slaughterhouses’ self-checking systems (SCSs) fail.
The aim of this study was to investigate the prerequisites for official control such as the functionality of the task distribution in meat inspection and certain meat inspection personnel-related factors. In addition, needs for improvement in slaughterhouses’ SCSs, meat inspection, and food safety inspections, including control measures used by the official veterinarians (OVs) and their efficacy, were examined. In the EU, competent authorities must ensure the quality of official control in slaughterhouses through internal or external audits, and the functionality of these audits was also studied.
Based on our results, meat inspection personnel (OVs and official auxiliaries [OAs]), slaughterhouse representatives, and officials in the central authority were mainly satisfied with the functionality of the present task distribution in meat inspection, although redistributing ante-mortem inspection from the OVs to the OAs was supported by some slaughterhouse representatives due to perceived economic benefit.
Ante-mortem inspection was assessed as the most important meat inspection task as a whole for meat safety, animal welfare, and prevention of transmissible animal diseases, and most of the respondents considered it important that the OVs perform antemortem inspection and whole-carcass condemnation in red meat slaughterhouses.
In a considerable number of slaughterhouses, OA or OV resources were not always sufficient and the lack of meat inspection personnel decreased the time used for food safety inspections according to the OVs, also affecting some of the red meat OAs’ post-mortem inspection tasks. The frequency with which OVs observed post-mortem inspection performed by the OAs varied markedly in red meat slaughterhouses. In addition, roughly one-third of the red meat OAs did not consider the guidance and support from the OVs to be adequate in post-mortem inspection.
According to our results, the most common non-compliance in slaughterhouses concerned hygiene such as cleanliness of premises and equipment, hygienic working methods, and maintenance of surfaces and equipment. Chief OVs in a few smaller slaughterhouses reported more frequent and severe non-compliances than other slaughterhouses, and in these slaughterhouses the usage of written time limits and enforcement measures by the OVs was more infrequent than in other slaughterhouses.
Deficiencies in documentation of food safety inspections and in systematic follow-up of corrections of slaughterhouses’ non-compliance had been observed in a considerable number of slaughterhouses. In meat inspection, deficiencies in inspection of the gastrointestinal tract and adjacent lymph nodes were most common and observed in numerous red meat slaughterhouses. Internal audits performed to evaluate the official control in slaughterhouses were considered necessary, and they induced correction of observed non-conformities. However, a majority of the interviewed OVs considered that the meat inspection should be more thoroughly audited, including differences in the rejections and their reasons between OAs. Auditors, for their part, raised a need for improved follow-up of the audits.
Our results do not give any strong incentive to redistribute meat inspection tasks between OVs, OAs, and slaughterhouse employees, although especially from the red meat slaughterhouse representatives’ point of view the cost efficiency ought to be improved. Sufficient meat inspection resources should be safeguarded in all slaughterhouses, and meat inspection personnel’s guidance and support must be emphasized when developing official control in slaughterhouses. OVs ought to focus on performing follow-up inspections of correction of slaughterhouses’ non-compliance systematically, and also the documentation of the food safety inspections should be developed.
Hygiene in slaughterhouses should receive more attention; especially in slaughterhouses with frequent and severe non-compliance, OVs should re-evaluate and intensify their enforcement.
The results attest to the importance of internal audits in slaughterhouses, but they could be developed by including auditing of the rejections and their underlying reasons and uniformity in meat inspection.