A common approach for ranking of microbiological and chemical hazards in foods based on risk assessment - useful but is it possible?

This article compares and contrasts microbial and chemical risk assessment methodologies in order to evaluate the potential for a common framework for ranking of risk of chemical and microbiological hazards, and developments needed for such a framework. An overview of microbial (MRA) and chemical (CRA) risk assessment is presented and important differences are highlighted. Two microbiological and two chemical hazard-food combinations were ranked based on both a margin of exposure and a risk assessment approach. The comparisons illustrated that it is possible to rank chemical and microbiological hazard-food combinations with traditional approaches from each domain and indicated that the rank order but not the absolute measures is similar using either approach. Including severity in the assessment using DALY reduced differences between hazards and affected the outcome more than which approach was used. Ranking frameworks should include assessment of uncertainty as an integral part of the ranking, and be based on assessment of risk, not safety, and expressed in a common health metric such as disease burden. Necessary simplifications to address data gaps can involve the use of default scenarios. Challenges include comparisons of case-based vs. non-case-based health-endpoints, e.g. biomarker concentration, and integration of the severity of health effects into ranking.

Salmonella risk to consumers via pork is related to the Salmonella prevalence in pig feed

Pigs are an important source of human infections with Salmonella, one of the most common causes of sporadic gastrointestinal infections and foodborne outbreaks in the European region. Feed has been estimated to be a significant source of Salmonella in piggeries in countries of a low Salmonella prevalence. To estimate Salmonella risk to consumers via the pork production chain, including feed production, a quantitative risk assessment model was constructed. The Salmonella prevalence in feeds and in animals was estimated to be generally low in Finland, but the relative importance of feed as a source of Salmonella in pigs was estimated as potentially high. Discontinuation of the present strict Salmonella control could increase the risk of Salmonella in slaughter pigs and consequent infections in consumers. The increased use of low risk and controlled feed ingredients could result in a consistently lower residual contamination in pigs and help the tracing and control of the sources of infections.

Shell egg handling and preparation practices in food service establishments in Finland

Foodborne outbreaks are often reported to be acquired at food service establishments. As a part of a quantitative risk assessment on the consumer risk of contracting Salmonella infection via shell eggs, we studied how small, medium, and large restaurants, institutional kitchens, and staff canteens (n=171) purchase, store, and use shell eggs. In addition, we estimated the fraction of raw and undercooked risky egg dishes among all egg dishes served in food service establishments of different sizes and types. The majority of establishments used shell eggs (78%), purchased eggs once per week (39%), and stored eggs at cool temperatures (82%). The size of the food service establishment had a less significant effect on shell egg preparation and handling practices than the type of the establishment. In particular, restaurants and institutional kitchens differed from each other. Restaurants purchased shell eggs more frequently, were more likely to store them at room temperature, stored shell eggs for a shorter period, and were more likely to prepare undercooked egg dishes than institutional kitchens. It was predicted that 6 to 20% of all different egg dishes prepared in a single randomly chosen food service establishment would be risky egg dishes with a 95% Bayesian credible interval of 0 to 96%, showing uncertainty because of the variability between kitchens and uncertainty in kitchen type-specific parameters. The results indicate that although most Finnish food service establishments had safe egg handling practices, a substantial minority expressed risky behavior. Compared with the egg consumption patterns in private Finnish households, however, practices in food service establishments did not prove to be more prone to risk.

Salmonella risk in imported fresh beef, beef preparations, and beef products

Additional guarantees (AGs) for Salmonella in imported defined animal-derived foods were agreed on for Finland when it was admitted to the European Community. The aim of this project was to evaluate the impact of these AGs on the prevalence of Salmonella in the Finnish beef supply and the adequacy of their scope. According to the quantitative Bayesian model, the efficacy of AGs was mainly dependent on the proportions of different beef categories imported and the true prevalence in the countries of origin. According to the model, AGs were able to reach their target in the referred year 1999 and kept the true Salmonella prevalence of beef imports below 1% with quantified uncertainty. The extension of AGs to all imported fresh beef would have reduced the Salmonella prevalence of beef imports from three- to fourfold, whereas expanding the implementation of AGs to all imports of fresh beef, beef preparations, and beef products would have resulted in a sixfold decrease. If current AGs targeting fresh beef intended to be sold as fresh or to be processed by the Finnish industry with processes not achieving 70 degrees C were not implemented, the 95% credible interval of Salmonella prevalence in the Finnish beef supply would be 0.2 to 1.3% (mean, 0.6%) instead of 0.1 to 1.2% (mean, 0.5%). However, if the prevalence in the exporting countries were to rise or the main import countries and/or magnitudes were to change, AGs would be of greater importance.

A quantitative risk assessment of the public health impact of the Finnish Salmonella control program for broilers

In order to study the public health effects of the Finnish Salmonella control program (FSCP), a quantitative risk assessment model of Salmonella from slaughtered broiler flocks to consumers was developed. Based on the model, approximately 0.21% of domestically produced broiler meat mass was contaminated with Salmonella (95% probability interval 0.05-0.48%). This model was combined to the model on primary production of broilers. By this way, the effect of eliminating breeder flocks from production which have tested positive for Salmonella and heat-treating the meat of detected positive broiler flocks on public health could be simulated. Based on the whole model, if detected positive breeder flocks were not removed this would result in 1.0-2.5 more reported human cases compared to the expected number of cases under current FSCP (95% predictive interval). Without heat treatment of meat the increase would be 2.9-5.4-fold and without both interventions 3.8-9.0-fold. In scenarios with one grandparent or five parent flocks infected, the combined effect of these two interventions was 9.3-25.8-fold and 4.9-11.7-fold compared to the baseline level under each scenario, respectively. The scenario analyses suggest that with a higher infection level, inclusion of both interventions will be more effective than either of the interventions alone. Replacement of half of the current retail broiler meat by meat with 20-40% contamination could result in 33-93 times more human cases compared to the expected value under current situations. On the basis of the model, the interventions applied in FSCP clearly protect the public health.

Predicting meningococcal disease outbreaks in structured populations

Rational decision making on whether some form of intervention would be necessary to control the spread of a meningococcal epidemic is based on predictions concerning its potential natural progression. Unfortunately, reliable predictions are difficult to make during the early stages of an outbreak. A stochastic discrete time epidemic model was applied to adaptively predict the development of outbreaks of meningococcal disease in 'closed' populations such as military garrisons or boarding schools, which are further divided into subgroups called 'units'. The performance of the adaptive method was assessed by using 3 simulated epidemics representing substantially different realizations in a 'garrison' of 20 units, with 68 men in each. Predictions of the weekly number of disease cases, of the number of carriers, and of the number of new infections were computed. Simulations suggest that predictions based only on the observed numbers of disease cases are generally inaccurate. These predictions can be improved if temporal observations on asymptomatic carriers in different units are utilized together with observed time series of the disease. A sample of 15 per cent from all units can be sufficient for a major improvement if the alternative is to obtain a full sample of only some units. Exploiting fully such information requires computer intensive Markov chain Monte Carlo methods.

The incidence of type 1 diabetes among children in Finland—rural–urban difference

The aim of the study was to investigate the incidence of type 1 diabetes among children aged 14 years or under according to the level of urbanization of the place of residence of children at the time of diagnosis in Finland during 1987 to 1996. The analysis was carried out using a Bayesian approach and GIS. The incidence was the highest in the rural heartland areas while the increase in incidence was sharpest in urban areas. The level of urbanization seems to explain only a part of the spatial variation in the incidence in Finland. It is possible that some environmental risk factors for type 1 diabetes have been more prevalent in rural heartland areas than in the rest of the country. These factors might have increased in urban environments in Finland particularly during the first half of 1990s.

Poliovirus surveillance by examining sewage water specimens: studies on detection probability using simulation models

Efficiency of environmental surveillance of poliovirus circulation was studied using simulation models. First, three transmission models were defined for describing different scenarios of poliovirus infections in a large unstructured population. Second, environmental factors, such as the total volume of the sewage network and losses of viruses, were modeled for computing the virus output at the sewage sampling site. Third, the effect of sampling and laboratory procedures was accounted for in the probability of detection, given the amount of polioviruses in a specimen. The simulation model can be used for theoretical assessments of the likely efficiency of environmental surveillance, compared with acute flaccid paralysis (AFP) surveillance. Under reasonable assumptions in a vaccinated population, the AFP surveillance can be outperformed if the poliovirus outbreak is not large. However, this depends on the assumed case-to-infection ratio and on the sampling frequency of the sewage water specimens. Increasing the latter will lead to a higher detection probability, which will further enhance the method based on environmental surveillance.

Bayesian analysis of geographical variation in the incidence of Type I diabetes in Finland

AIMS/HYPOTHESIS

In Finland, the incidence of Type I (insulin-dependent) diabetes mellitus among children aged 14 years or under is the highest in the world. The increase in incidence is approximately 3% per year. A marked geographical variation in incidence was reported in Finland during the late 1980s. Our aim was to explore the most recent regional pattern in incidence of Type I diabetes in Finland.

METHODS

Data on the nationwide incidence of childhood diabetes in Finland was obtained from the Prospective Childhood Diabetes Registry for the periods 1987-1991 and 1992-1996. Population data was obtained from the National Population Registry. The geographical pattern of incidence was studied applying a Bayesian hierarchical approach and Geographical Information Systems. The inferences from the data was based on the estimated geographical intensity of diabetes.

RESULTS

There was a clear evidence of geographic variation for the risk of childhood diabetes during the entire 10-year period. The high-risk areas were found in the wide belt crossing the central part of Finland. Comparison of the estimated intensity of diabetes between the two 5-year periods showed that the geographical pattern of diabetes risk has changed over time. Our analyses also confirmed the existence of a few persistent high-risk and low-risk areas in Finland.

CONCLUSION/INTERPRETATION

The finding of high-risk areas of childhood Type I diabetes suggests that specific genetic or environmental risk factors have become greater in certain geographic locations in Finland.

Small area estimation of incidence of cancer around a known source of exposure with fine resolution data

OBJECTIVES

To describe the small area system developed in Finland. To illustrate the use of the system with analyses of incidence of lung cancer around an asbestos mine. To compare the performance of different spatial statistical models when applied to sparse data.

METHODS

In the small area system, cancer and population data are available by sex, age, and socioeconomic status in adjacent "pixels", squares of size 0.5 km x 0.5 km. The study area was partitioned into sub-areas based on estimated exposure. The original data at the pixel level were used in a spatial random field model. For comparison, standardised incidence ratios were estimated, and full bayesian and empirical bayesian models were fitted to aggregated data. Incidence of lung cancer around a former asbestos mine was used as an illustration.

RESULTS

The spatial random field model, which has been used in former small area studies, did not converge with present fine resolution data. The number of neighbouring pixels used in smoothing had to be enlarged, and informative distributions for hyperparameters were used to stabilise the unobserved random field. The ordered spatial random field model gave lower estimates than the Poisson model. When one of the three effects of area were fixed, the model gave similar estimates with a narrower interval than the Poisson model.

CONCLUSIONS

The use of fine resolution data and socioeconomic status as a means of controlling for confounding related to lifestyle is useful when estimating risk of cancer around point sources. However, better statistical methods are needed for spatial modelling of fine resolution data.

Probabilistic small area risk assessment using GIS-based data: a case study on Finnish childhood diabetes. Geographic information systems

A Bayesian hierarchical spatial model is constructed to describe the regional incidence of insulin dependent diabetes mellitus (IDDM) among the under 15-year-olds in Finland. The model exploits aggregated pixel-wise locations for both the cases and the population at risk. Typically such data arise from combining geographic information systems (GIS) with large databases. The dates of diagnosis and locations of the cases are observed from 1987 to 1996. The population at risk counts are available for every second year during the same period. A hierarchical model is suggested for the pixel wise case counts, including a population model to account for the uncertainty of the population at risk over the years. The model is applied in the construction of disease maps (aggregated 100 km(2) pixels), and spatial posterior predictive distributions are computed to study whether there can be found a statistically exceptional number of cases in a small area of interest.

Probabilistic small area risk assessment using GIS-based data: a case study on Finnish childhood diabetes. Geographic information systems

A Bayesian hierarchical spatial model is constructed to describe the regional incidence of insulin dependent diabetes mellitus (IDDM) among the under 15-year-olds in Finland. The model exploits aggregated pixel-wise locations for both the cases and the population at risk. Typically such data arise from combining geographic information systems (GIS) with large databases. The dates of diagnosis and locations of the cases are observed from 1987 to 1996. The population at risk counts are available for every second year during the same period. A hierarchical model is suggested for the pixel wise case counts, including a population model to account for the uncertainty of the population at risk over the years. The model is applied in the construction of disease maps (aggregated 100 km(2) pixels), and spatial posterior predictive distributions are computed to study whether there can be found a statistically exceptional number of cases in a small area of interest.

Predicting the course of meningococcal disease outbreaks in closed subpopulations

A stochastic epidemic model was applied to meningococcal disease outbreaks in defined small populations such as military garrisons and schools. Meningococci are spread primarily by asymptomatic carriers and only a small proportion of those infected develop invasive disease. Bayesian predictions of numbers of invasive cases were developed, based on observed data using a stochastic epidemic model. We used additional data sets to model both disease probability and duration of carriage. Markov chain Monte Carlo sampling techniques were used to compute the full posterior distribution which summarized all information drawn together from multiple sources.

Detection of overall space-time clustering in a non-uniformly distributed population. DiMe Study Group

We developed a test statistic based on an approach of Whittemore et al. (1987) to detect space-time clustering for non-infectious diseases. We extended the spatial test of Whittemore et al. by deriving conditional probabilities for Poisson distributed random variables. To combine spatial and time distances we defined a distance matrix D, where dij is the distance between the ith and jth cell in a three-dimensional space-time grid. Spatial and temporal components are controlled by a weight. By altering the weight, both marginal tests and the intermediate test can be reached. Allowing a continuum from a pure spatial to a pure temporal test, the best result will be gained by trying different weights, because the occurrence of a disease might show some temporal and some spatial tendency to cluster. We examined the behaviour of the test statistic by simulating different distributions for cases and the population. The test was applied to the incidence data of insulin-dependent diabetes mellitus in Finland. This test could be used in the analysis of data which are localized according to map co-ordinates, by addresses or postcodes. This information is important when using the Geographical Information System (GIS) technology to compute the pairwise distances needed for the proposed test.

A statistical model of transmission of Hib bacteria in a family

The simultaneous estimation of family and community transmission rates as well as cure rates from panel data in a recurrent Hib (Haemophilus influenzae type b bacteria) infection is considered. An individual-based stationary Markov process model with constant hazards in two age groups is applied to describe recurrent asymptomatic Hib infection in a family with small children. The problem of estimation is solved in terms of the Bayesian posterior of the model parameters. The model is used to predict prevalence and incidence of Hib carriage in families as a function of the family size and age structure.

Comparative double-blind study of flupenthixol decanoate and fluphenazine decanoate in the treatment of patients relapsing in a schizophrenic symptomatology

Thirty-two chronic schizophrenics who had relapsed entered a double-blind randomised study and were followed-up for 2 years with the intention of measuring any difference in therapeutic effect and side effects between flupenthixol decanoate and fluphenazine decanoate. No differences could be seen as regards the global effect or the effect on the schizophrenic symptomatology during the first 6 months. After 1 year of treatment flupenthixol decanoate showed a trend towards a better effect on schizophrenic symptomatology. A corresponding result was seen for the depressive symptoms. There were no differences in the appearance of side effects. The need for additional neuroleptics in the initial phase seemed to be identical for both drugs. A possible slow antipsychotic effect with flupenthixol decanoate is probably due to the administered dose being somewhat low (in the present study approximately 31 mg flupenthixol corresponding to 27 mg fluphenazine). This suggests that flupenthixol should have been given in a somewhat higher dose (25 mg fluphenazine decanoate corresponding to 40 mg flupenthixol decanoate).