Association between food control inspection grades and regional incidence of infectious foodborne diseases in Finland

ABSTRACT We studied regional food control inspection grades and their relation to regional incidence of domestically acquired foodborne diseases (caused by Campylobacter spp. Salmonella spp. enterohemorrhagic Escherichia coli (EHEC), and Listeria monocytogenes) using food control inspection data of local food business operators and infectious disease data from 2014 to 2019 from Finland. We observed that inferior overall inspection grades were associated with increased incidence of Salmonella infections (p=0.02). Specifically, inferior grades on cleanliness of facilities, surfaces, and equipment were associated with increased incidence of Salmonella infections (p=0.04). For this topical inspection area, a high effect size was also seen for Campylobacter infections (p=0.06). Of the individual inspection items, an association between increased incidence of Campylobacter infections and inferior grades on storage of foodstuffs (p=0.01) and verification of hygiene proficiency (p=0.03) was observed. These results suggest that food control recognizes non-compliances that may predispose to foodborne diseases.


Introduction
The major aim of food control is prevention of foodborne diseases.During the past two decades, the annual number of reported foodborne outbreaks in Finland ranged from 22 to 73, affecting 500 to 1600 people (FFA 2022b).In the years before the COVID-19 pandemic, the average number of infections caused by foodborne pathogens reported to the National Infectious Diseases register has been slightly above 10,000 infections per year (THL 2022).It is known that foodborne diseases are underreported, since only a minor proportion of patients seek medical attention as symptoms are usually self-limiting.Even when medical attention is sought, a stool sample may not be taken, preventing identification of the causative agent.Thus, the true numbers of foodborne infections are considerably higher than those reported.For example, Haagsma et al. (2013) concluded that the reported incidence of Campylobacter is between 1% and 11% of the true incidence, depending on the country.Thus, foodborne diseases can be considered a significant public health issue.
To prevent foodborne infections, multiple measures should be implemented when handling food.These measures include, for example, prevention of cross-contamination, proper heating, temperature management, and personnel hygiene.These items are assessed during food control inspections.The most important preventive measures are pathogen-specific.For example, for Campylobacter, heat treatment, and prevention of cross-contamination are important (Altekruse et al. 1999).Whereas prevention of Salmonella requires heat treatment, proper temperature management, prevention of cross-contamination, good hand hygiene (Ehuwa et al. 2021), and abstaining from work if gastrointestinal symptoms occur (Hundy and Cameron 2002).As sources of different pathogens vary (Kumagai et al. 2020;Mughini-Gras et al. 2021;Chanamé Pinedo et al. 2022), it is also important to understand which foods impose greatest risks for each pathogen so that necessary preventive measures can be applied.
Restaurants are a significant source of foodborne diseases (Jones et al. 2006;Domingues et al. 2012).According to the European Food Safety Authority (EFSA) and European Centre for Disease Prevention and Control (ECDC), 40% of foodborne outbreaks originate from food service establishments (EFSA and CDC 2021).Eating at a food service establishment exposes the consumer to the food safety behavior of the food handlers.In the UK, various types of misconduct among food handlers have been reported (Jones et al. 2017).For example, the ability to prepare food safely can be compromised by environment or equipment not suitable for operation, time pressure, or poor education and training (Green and Selman 2005).The effect of non-compliance on food safety in a restaurant is typically amplified, as multiple persons are usually exposed to the food safety hazard caused by the non-compliance.Other common sources of foodborne diseases include catering facilities, institutional kitchens, and domestic kitchens (CDC 2019).Foodborne outbreaks may also originate from contaminated raw material or food production establishments (Jacks et al. 2016;Huusko et al. 2017).These outbreaks typically spread widely geographically and affect more people than local outbreaks (Dewey-Mattia et al. 2018;EFSA and CDC 2021).
The association between inspection grades and foodborne diseases has typically been studied by comparing outbreak and non-outbreak restaurants.The results of these studies have been contradictory.In Finland, Leinonen et al. (2022) found differences in food control inspection grades regarding order and cleanliness between outbreak and non-outbreak institutional catering establishments.Irwin et al. (1989) and Fleetwood et al. (2019) found that outbreak restaurants had inferior inspection results compared to non-outbreak restaurants.In another study, an association between increased risk of being an outbreak restaurant and lower overall inspection grade was also observed (Buchholz et al. 2002).In addition, several inspection items were related to outbreak restaurants.Petran et al. (2012) found that there were significantly more violations at outbreak restaurants.On the other hand, several studies have not found an association between inspection scores and foodborne outbreaks (Cruz et al. 2001;Jones et al. 2004;Lee and Hedberg 2016).These studies provide valuable information on the characteristics of outbreak restaurants.However, they are restricted to foodborne outbreaks and do not consider sporadic infections or infections that are not recognized as part of outbreaks.Therefore, studies that consider the available data, including sporadic infections and inspections, are warranted to identify the associations between inspection results and incidences of foodborne diseases.
Our aim was to study the association between regional food control inspection grades and the regional incidence of infections caused by Campylobacter spp., Salmonella spp., Enterohemorrhagic Escherichia coli (EHEC), and Listeria monocytogenes.Our hypothesis was that inferior inspection grades are associated with an increased incidence of foodborne diseases.The results increase the understanding on how the inspected issues and associated compliance are linked to the incidence of foodborne diseases, which can then be used to develop the effectiveness of inspections.To our knowledge, this is the first study to explore regional food control inspection grades and their association with the incidence of foodborne diseases.

Inspection reports and infectious disease data
Clinical microbiology laboratories in Finland are required to report laboratory-confirmed infectious disease cases to the Finnish Infectious Disease Register (FIDR) maintained by Finnish Institute for Health and Welfare.Materials for this study included FIDR data on foodborne pathogen infections from 2014 to 2019 and food control inspection reports from all 62 food control units in Finland from 2014 to 2019.In addition, FIDR data from 2020 were used for imputation of missing infection origin data.As we examined regional inspection grades and their association to incidence of domestically acquired foodborne diseases, we included only local food business operators (FBOs) in our analyses and disregarded food production establishments that typically distribute their products more widely.We considered restaurants, institutional kitchens, retail stores, pubs, coffee shops, and outdoor food sale premises as local FBOs.Inspection data were obtained from the Finnish Food Authority, which acts as the central agency directing food control in Finland.Altogether, there were 119 469 food control inspections included in the analyses.
In the Finnish food control inspection system, inspections are graded based on a four-level scale (A = excellent, complies with requirements; B = good, small issues that do not impair food safety or mislead customers; C = to be corrected, issues that impair food safety or mislead consumers; D = poor, issues that jeopardize food safety or considerably mislead consumers, or FBO has not corrected non-compliances previously noted) (FFA 2021).The overall inspection grade is determined by the lowest individual inspection grade.Here we define overall grades B, C, and D as noncompliant.If the overall score is C or D, follow-up inspection is performed to ensure that noncompliances are corrected.Grades were converted to numeric scale (A = 4, B = 3, C = 2, D = 1) when average grades were calculated.There are 17 topical inspection areas that consist altogether of 66 individual inspection items, out of which only some are covered at each inspection.In our analyses, we included inspection items that were inspected at least 10 000 times during the study period.Inspections are carried out risk based; higher the risk, more frequent inspections.Based on risk rating, food service establishments and food retailers are inspected 0.35-3 times a year (FFA 2022a).Follow-up inspections may increase the inspection frequency.
Foodborne pathogens included in the FIDR are Campylobacter spp., Salmonella spp., Yersinia spp., EHEC, Shigella spp., Listeria monocytogenes, Giardia spp., Cryptosporidium spp., and Norovirus.As data regarding country of origin of infections were incomplete, Campylobacter and Salmonella were studied further as their data enabled imputation of the infection origin.For other pathogens, the overall number of infections was insufficient or origin of infections was missing for too large proportion of infections for imputation or a meaningful interpretation of statistical testing.However, as EHEC and Listeria only had a low number of total infections, but not many infections with missing infection origin, we included these pathogens for visual examination but abstained from statistical testing.

Statistical analysis
We performed multiple imputation for missing infection origin information for Campylobacter and Salmonella infections.Imputation was performed due to differences in the domestic and foreign infection ratios and in the percentage of cases with missing information on infection origin between food control units, which would have biased the estimates.For imputation, we utilized infection data from April -December 2020, when incidence of both domestic and foreign infections were affected by restrictions that the COVID-19 pandemic had on traveling and on FBOs.This allowed us to estimate the proportion of infections with missing origin of infection that were to be imputed as domestic.A hundred imputed datasets were generated for both Campylobacter and Salmonella.Analyses were performed for each data set, and final results were pooled from these analyses.For Campylobacter, the percentage of infections with missing infection origin was higher for food control units with inferior grades.This led to more observations being imputed to food control units with low grades.Imputation methodology is described in detail in Supplementary material.
The incidence of foodborne diseases and average inspection grades for local FBOs were calculated for the areas of the 62 food control inspection units.Based on the normality of residuals, either linear regression or robust regression (Huber 1973) was used to determine the association between the incidence of foodborne diseases and 1) average overall inspection grades, 2) average grades of topical inspection areas, and 3) average grades of individual inspection items.Shapiro-Wilk test was used to assess normality.In accordance with Lundén et al. (2021), we identified the mean income of the area and percentage of people living in urban areas as confounding factors and included them in the analyses to control for them.Data regarding income and percentage of people living in urban areas were retrieved from Statistics Finland (Statistics Finland 2022).Analyses were performed for each imputed dataset and results were pooled using SAS PROC MIANALYZE.
Statistical significance was determined as p < 0.05.Analyses were conducted using SAS 9.4 (SAS Institute, Cary NC).Figures were created using SAS 9.4 and R version 4.0.3(R Core Team 2020).

Incidence of foodborne diseases
Campylobacter was the most common pathogen causing foodborne disease in 2014-2019.There were on average 4658 reported infections per year, including domestic and foreign infections, and infections for which the country of origin was unknown (Table 1).After imputation, there were on average 2290 domestic Campylobacter infections per year.There were 1495 reported yearly infections for Salmonella (Table 1).After imputation, there were on average 323 domestic Salmonella infections per year.The reported average yearly number of domestically acquired EHEC and Listeria infections were 71 and 52, respectively.

Grade distribution of inspections
Of the inspections, 43.2% received overall grade A, 43.4% grade B, 12.7% grade C, and 0.7% grade D (Table 2).There were considerable differences on grade distributions of individual inspection items.The item with the most non-compliances ("Chilling") had grade A in only 68.2% of inspections; the inspection item with the least non-compliances ("Separation of product groups and hygiene during selling and serving") had grade A in 97.9% of inspections (Table 2).
The average overall grades (converted to numeric) of food control units ranged from 2.62 to 3.77 (Table 2).For individual inspection items, the widest ranges were on items "General labeling" (range 3.08-3.97)and "Chilling" (range 3.14-3.96).The lowest variation was on inspection items "Separation of product groups and hygiene during selling and serving" (range 3.85-4.00)and "Recalls" (range 3.83-4.00).

Association between grades of topical areas and individual inspection items with Salmonella and Campylobacter incidence
The incidence of Salmonella infections decreased when the inspection grades for cleanliness of facilities, surfaces, and equipment improved (robust regression, B = −4.63[95% CI: −9.07; −0.18], p = 0.04) (Table 3).For Campylobacter, a high effect size was also seen on the same topical inspection area (linear regression, B = −16.53[95% CI: −34.05; 0.98], p = 0.06) (Table 3).A decline in Campylobacter incidence was seen when inspection grades of storage of foodstuffs (linear regression, p = 0.01) and verification of hygiene proficiency (linear regression, p = 0.03) improved (Figure 2).For Salmonella, there were no individual inspection items with statistically significant association (Figure 3)

Discussion
This study highlights the overall relevance of food control inspections to food safety.The results show that inspection grades vary between food control units, and this appears to reflect the number of foodborne diseases.Such results provide evidence that routine surveillance of the inspection data could be used to improve food safety as suggested by Firestone et al. (2021).We found an association between regional overall inspection grades and incidence of Salmonella infections.
Inferior grades were associated with increased incidence.For both Salmonella and Campylobacter, we also identified associations between disease incidence and topical inspection areas or individual inspection items.The associations indicate that food control is focused on relevant issues, and additional measures to correct non-compliances could improve food safety.This would be especially important in food control units where inspection grades are below average.The grades of some inspected items were significantly associated with the incidence of Salmonella and Campylobacter infection.For Salmonella, an association was found between incidence and inspection grades concerning "Cleanliness of facilities, surfaces, and equipment".Non-compliances in cleanliness were identified as a risk factor for Salmonella infections in previous studies (Patel et al. 2010;Osimani et al. 2016;Appling et al. 2018).In our study, cleanliness also exhibited a high effect size between inspection grades and Campylobacter incidence.These findings are not surprising, as Salmonella and Campylobacter can survive for long periods on food contact surfaces (De Cesare et al. 2003) and can also form biofilms (Lamas et al. 2018).Thus, poor cleaning creates a significant risk for cross-contamination.Cleanliness is a central theme of food safety and food control inspections.Based on these results, the identified non-compliances in cleanliness are relevant for food safety; more effective interventions on these non-compliances may reduce the incidence of foodborne diseases.
In our study, Campylobacter incidence declined when inspection grades for "Storage of foodstuffs" improved.This inspection item includes, for example, appropriateness of storage facilities and prevention of cross-contamination during storage.Storage of foodstuffs has been previously identified as an important food safety concern (Buchholz et al. 2002;Gormley et al. 2011;Zhao et al. 2022).Another individual inspection item with an association between inspection grades and Campylobacter incidence was "Verification of hygiene proficiency", which refers to the hygiene passport test used in Finland.Preparation (i.e.study or training) for the test increases the food safety knowledge level, especially if initial knowledge is poor (Vaarala et al. 2021).Our results indicate that the hygiene passport is relevant in the prevention of foodborne diseases, and it is important to verify at inspections that the food handlers have completed the test.
Visual examination of scatterplots revealed that also EHEC incidence exhibited a declining trend as regional inspection grades improved.The scatterplot for Listeria did not reveal a similar association.This may be because risk products for Listeria infection include ready-to-eat products that are often produced in food production establishments and are distributed widely geographically (Nakari et al. 2014;Jacks et al. 2016).Thus, infections may also occur in areas other than where Campylobacter incidence during the study period was 41.7 infections/100 000 persons.
b Salmonella incidence during the study period was 5.9 infections/100 000 persons.
the establishment is located and from where the problem arises.Thus, it was expected that the association between inspection grades and Listeria incidence was not observed in this study.However, the results do not indicate that inspections would have no effect on the prevention of Listeria contamination in food establishments.In fact, food control has been shown to recognize non-compliance that predisposes to L. monocytogenes occurrence in fish processing plants (Aalto-Araneda et al. 2018), and accomplished control measures have resulted in a decrease of reported listeriosis cases (Nakari et al. 2014).This study describes an innovative approach to estimate the proportion of foodborne diseases that are domestically acquired using the information available on the change in incidence of foodborne diseases during the COVID-19 pandemic.The large proportion of cases with unknown origin has previously hampered analyses concerning domestic cases.However, the restrictions implemented due to the COVID-19 pandemic have reduced the incidence of both domestic and travel-associated foodborne diseases.For Campylobacter and Salmonella, differences in the reduction of incidences enabled us to estimate the proportion of domestic infections among those whose origin of infection was unknown.This in turn enabled multiple imputation to estimate the incidence in different food control units and to perform analyses regarding associations between regional incidences and inspection grades.
Although imputation increases the reliability of our estimates of the regional incidence of domestic infections, it also adds another layer of potential bias to estimates of which effect we cannot appraise.In our study, imputation was necessary since there are major differences in domestic and foreign infection ratios and in the percentage of cases with missing information on infection origin between food control units.Due to limitations of the data, imputation could not be implemented for pathogens other than Campylobacter and Salmonella, and we were not able to include these pathogens comprehensively in the study.There are also other limitations to this study that should be addressed.First, the incidence of reported foodborne diseases do not give a complete picture of incidence, as a large portion of infections remain unreported as the diseases are usually self-limiting and infected persons do not seek medical care.Also, although majority of infections caused by these pathogens are foodborne, there are also other transmission pathways (Beshearse et al. 2021).However, it is a reasonable assumption that there are no considerable differences in unreported cases between food control units, nor in the proportion of infections that are foodborne, and thus these results can be generalized to the overall incidence.Second, we cannot draw conclusions about causality based on this study; this would require a different study design, such as a retrospective case-control study comparing FBOs from which foodborne diseases originated and those from which there were not.

Figure 1 .
Figure 1.Association between average overall inspection grades and incidence of foodborne pathogens.Each dot represents a food control unit.Black dots represent confirmed domestic data.For Campylobacter and Salmonella, gray dots represent data after imputation.

RecallsFigure 2 .
Figure 2. Coefficient plot for association of individual inspection items with Campylobacter incidence.Regression coefficient and 95% confidence interval are shown.Negative regression coefficient indicates that inferior average grades are associated with increased incidence.

Table 1 .
Average yearly number of infections caused by foodborne pathogens in Finland from 2014 to 2019.
a Mean yearly incidence per 100 000 persons.

Table 2 .
Grade distribution of topical inspection areas and individual inspection items in Finland from 2014 to 2019.
a Range of average grades of food control units; A = 4, B = 3, C = 2, and D = 1.

Table 3 .
Association between average inspection grades of topical inspection areas and incidence of Campylobacter and Salmonella infections from 2014 to 2019.Pooled results of linear/robust regressions from 100 imputed datasets.