Critical comparison of statistical methods for quantifying variability and uncertainty of microbial responses from experimental data.

Variability and uncertainty are important factors for quantitative microbiological risk assessment (QMRA). In this context, variability refers to inherent sources of variation, whereas uncertainty refers to imprecise knowledge or lack of it. In this work we compare three statistical methods to estimate variability in the kinetic parameters of microbial populations: mixed-effect models, multilevel Bayesian models, and a simplified algebraic method previously suggested. We use two case studies that analyse the influence of three levels of variability: (1) between-strain variability (different strains of the same species), (2) within-strain variability (biologically independent reproductions of the same strain) and, at the most nested level, (3) experimental variability (species independent technical lab variability resulting in uncertainty about the population characteristic of interest) on the growth and inactivation of Listeria monocytogenes. We demonstrate that the algebraic method, although relatively easy to use, overestimates the contribution of between-strain and within-strain variability due to the propagation of experimental variability in the nested experimental design. The magnitude of the bias is proportional to the variance of the lower levels and inversely proportional to the number of repetitions. This bias was very relevant in the case study related to growth, whereas for the case study on inactivation the resulting insights in variability were practically independent of the method used. The mixed-effects model and the multilevel Bayesian models calculate unbiased estimates for all levels of variability in all the cases tested. Consequently, we recommend using the algebraic method for initial screenings due to its simplicity. However, to obtain parameter estimates for QMRA, the more complex methods should generally be used to obtain unbiased estimates.

Rise and fall of the new variant of Chlamydia trachomatis in Sweden: mathematical modelling study.

OBJECTIVES: A new variant of Chlamydia trachomatis (nvCT) was discovered in Sweden in 2006. The nvCT has a plasmid deletion, which escaped detection by two nucleic acid amplification tests (Abbott-Roche, AR), which were used in 14 of 21 Swedish counties. The objectives of this study were to assess when and where nvCT emerged in Sweden, the proportion of nvCT in each county and the role of a potential fitness difference between nvCT and co-circulating wild-type strains (wtCT). METHODS: We used a compartmental mathematical model describing the spatial and temporal spread of nvCT and wtCT. We parameterised the model using sexual behaviour data and Swedish spatial and demographic data. We used Bayesian inference to fit the model to surveillance data about reported diagnoses of chlamydia infection in each county and data from four counties that assessed the proportion of nvCT in multiple years. RESULTS: Model results indicated that nvCT emerged in central Sweden (Dalarna, Gävleborg, Västernorrland), reaching a proportion of 1% of prevalent CT infections in late 2002 or early 2003. The diagnostic selective advantage enabled rapid spread of nvCT in the presence of high treatment rates. After detection, the proportion of nvCT decreased from 30%-70% in AR counties and 5%-20% in counties that Becton Dickinson tests, to around 5% in 2015 in all counties. The decrease in nvCT was consistent with an estimated fitness cost of around 5% in transmissibility or 17% reduction in infectious duration. CONCLUSIONS: We reconstructed the course of a natural experiment in which a mutant strain of C. trachomatis spread across Sweden. Our modelling study provides support, for the first time, of a reduced transmissibility or infectious duration of nvCT. This mathematical model improved our understanding of the first nvCT epidemic in Sweden and can be adapted to investigate the impact of future diagnostic escape mutants.

Mismatch Amplification Mutation Assay-Based Real-Time PCR for Rapid Detection of Neisseria gonorrhoeae and Antimicrobial Resistance Determinants in Clinical Specimens.

Molecular methods are often used for Neisseria gonorrhoeae detection, but complete definition of antimicrobial resistance (AMR) patterns still requires phenotypic tests. We developed an assay that both identifies N. gonorrhoeae and detects AMR determinants in clinical specimens. We designed a mismatch amplification mutation assay (MAMA)-based SYBR green real-time PCR targeting one N. gonorrhoeae-specific region (opa); mosaic penA alleles (Asp345 deletion [Asp345del], Gly545Ser) associated with decreased susceptibility to cephalosporins; and alterations conferring resistance to ciprofloxacin (GyrA Ser91Phe), azithromycin (23S rRNA A2059G and C2611T), and spectinomycin (16S rRNA C1192T). We applied the real-time PCR to 489 clinical specimens, of which 94 had paired culture isolates, and evaluated its performance by comparison with the performance of commercial diagnostic molecular and phenotypic tests. Our assay exhibited a sensitivity/specificity of 93%/100%, 96%/85%, 90%/91%, 100%/100%, and 100%/90% for the detection of N. gonorrhoeae directly from urethral, rectal, pharyngeal, cervical, and vaginal samples, respectively. The MAMA strategy allowed the detection of AMR mutations by comparing cycle threshold values with the results of the reference opa reaction. The method accurately predicted the phenotype of resistance to four antibiotic classes, as determined by comparison with the MIC values obtained from 94 paired cultures (sensitivity/specificity for cephalosporins, azithromycin, ciprofloxacin, and spectinomycin resistance, 100%/95%, 100%/100%, 100%/100%, and not applicable [NA]/100%, respectively, in genital specimens and NA/72%, NA/98%, 100%/97%, and NA/96%, respectively, in extragenital specimens). False-positive results, particularly for the penA Asp345del reaction, were observed predominantly in pharyngeal specimens. Our real-time PCR assay is a promising rapid method to identify N. gonorrhoeae and predict AMR directly in genital specimens, but further optimization for extragenital specimens is needed.

Age difference between heterosexual partners in Britain: Implications for the spread of Chlamydia trachomatis.

Heterosexual partners often differ in age. Integrating realistic patterns of sexual mixing by age into dynamic transmission models has been challenging. The effects of these patterns on the transmission of sexually transmitted infections (STI) including Chlamydia trachomatis (chlamydia), the most common bacterial STI are not well understood. We describe age mixing between new heterosexual partners using age- and sex-specific data about sexual behavior reported by people aged 16-63 years in the 2000 and 2010 British National Surveys of Sexual Attitudes and Lifestyles. We incorporate mixing patterns into a compartmental transmission model fitted to age- and sex-specific, chlamydia positivity from the same surveys, to investigate C. trachomatis transmission. We show that distributions of ages of new sex partners reported by women and by men in Britain are not consistent with each other. After balancing these distributions, new heterosexual partnerships tend to involve men who are older than women (median age difference 2, IQR -1, 5 years). We identified the most likely age combinations of heterosexual partners where incident C. trachomatis infections are generated. The model results show that in >50% of chlamydia transmitting partnerships, at least one partner is ≥25 years old. This study illustrates how sexual behavior data can be used to reconstruct detailed sexual mixing patterns by age, and how these patterns can be integrated into dynamic transmission models. The proposed framework can be extended to study the effects of age-dependent transmission on incidence in any STI.

Molecular relatedness of ESBL/AmpC-producing Escherichia coli from humans, animals, food and the environment: a pooled analysis.

Background: In recent years, ESBL/AmpC-producing Escherichia coli (ESBL/AmpC-EC) have been isolated with increasing frequency from animals, food, environmental sources and humans. With incomplete and scattered evidence, the contribution to the human carriage burden from these reservoirs remains unclear. Objectives: To quantify molecular similarities between different reservoirs as a first step towards risk attribution. Methods: Pooled data on ESBL/AmpC-EC isolates were recovered from 35 studies in the Netherlands comprising >27 000 samples, mostly obtained between 2005 and 2015. Frequency distributions of ESBL/AmpC genes from 5808 isolates and replicons of ESBL/AmpC-carrying plasmids from 812 isolates were compared across 22 reservoirs through proportional similarity indices (PSIs) and principal component analyses (PCAs). Results: Predominant ESBL/AmpC genes were identified in each reservoir. PCAs and PSIs revealed close human-animal ESBL/AmpC gene similarity between human farming communities and their animals (broilers and pigs) (PSIs from 0.8 to 0.9). Isolates from people in the general population had higher similarities to those from human clinical settings, surface and sewage water and wild birds (0.7-0.8), while similarities to livestock or food reservoirs were lower (0.3-0.6). Based on rarefaction curves, people in the general population had more diversity in ESBL/AmpC genes and plasmid replicon types than those in other reservoirs. Conclusions: Our 'One Health' approach provides an integrated evaluation of the molecular relatedness of ESBL/AmpC-EC from numerous sources. The analysis showed distinguishable ESBL/AmpC-EC transmission cycles in different hosts and failed to demonstrate a close epidemiological linkage of ESBL/AmpC genes and plasmid replicon types between livestock farms and people in the general population.

Comparative Exposure Assessment of ESBL-Producing Escherichia coli through Meat Consumption.

The presence of extended-spectrum ß-lactamase (ESBL) and plasmidic AmpC (pAmpC) producing Escherichia coli (EEC) in food animals, especially broilers, has become a major public health concern. The aim of the present study was to quantify the EEC exposure of humans in The Netherlands through the consumption of meat from different food animals. Calculations were done with a simplified Quantitative Microbiological Risk Assessment (QMRA) model. The model took the effect of pre-retail processing, storage at the consumers home and preparation in the kitchen (cross-contamination and heating) on EEC numbers on/in the raw meat products into account. The contribution of beef products (78%) to the total EEC exposure of the Dutch population through the consumption of meat was much higher than for chicken (18%), pork (4.5%), veal (0.1%) and lamb (0%). After slaughter, chicken meat accounted for 97% of total EEC load on meat, but chicken meat experienced a relatively large effect of heating during food preparation. Exposure via consumption of filet americain (a minced beef product consumed raw) was predicted to be highest (61% of total EEC exposure), followed by chicken fillet (13%). It was estimated that only 18% of EEC exposure occurred via cross-contamination during preparation in the kitchen, which was the only route by which EEC survived for surface-contaminated products. Sensitivity analysis showed that model output is not sensitive for most parameters. However, EEC concentration on meat other than chicken meat was an important data gap. In conclusion, the model assessed that consumption of beef products led to a higher exposure to EEC than chicken products, although the prevalence of EEC on raw chicken meat was much higher than on beef. The (relative) risk of this exposure for public health is yet unknown given the lack of a modelling framework and of exposure studies for other potential transmission routes.

Practicalities of using non-local or non-recent multilocus sequence typing data for source attribution in space and time of human campylobacteriosis.

In this study, 1208 Campylobacter jejuni and C. coli isolates from humans and 400 isolates from chicken, collected in two separate periods over 12 years in The Netherlands, were typed using multilocus sequence typing (MLST). Statistical evidence was found for a shift of ST frequencies in human isolates over time. The human MLST data were also compared to published data from other countries to determine geographical variation. Because only MLST typed data from chicken, taken from the same time point and spatial location, were available in addition to the human data, MLST datasets for other Campylobacter reservoirs from selected countries were used. The selection was based on the degree of similarity of the human isolates between countries. The main aim of this study was to better understand the consequences of using non-local or non-recent MLST data for attributing domestically acquired human Campylobacter infections to specific sources of origin when applying the asymmetric island model for source attribution. In addition, a power-analysis was done to find the minimum number of source isolates needed to perform source attribution using an asymmetric island model. This study showed that using source data from other countries can have a significant biasing effect on the attribution results so it is important to carefully select data if the available local data lack in quality and/or quantity. Methods aimed at reducing this bias were proposed.

Risk factors for campylobacteriosis of chicken, ruminant, and environmental origin: a combined case-control and source attribution analysis.

BACKGROUND: Campylobacteriosis contributes strongly to the disease burden of food-borne pathogens. Case-control studies are limited in attributing human infections to the different reservoirs because they can only trace back to the points of exposure, which may not point to the original reservoirs because of cross-contamination. Human Campylobacter infections can be attributed to specific reservoirs by estimating the extent of subtype sharing between strains from humans and reservoirs using multilocus sequence typing (MLST). METHODOLOGY/PRINCIPAL FINDINGS: We investigated risk factors for human campylobacteriosis caused by Campylobacter strains attributed to different reservoirs. Sequence types (STs) were determined for 696 C. jejuni and 41 C. coli strains from endemic human cases included in a case-control study. The asymmetric island model, a population genetics approach for modeling Campylobacter evolution and transmission, attributed these cases to four putative animal reservoirs (chicken, cattle, sheep, pig) and to the environment (water, sand, wild birds) considered as a proxy for other unidentified reservoirs. Most cases were attributed to chicken (66%) and cattle (21%), identified as the main reservoirs in The Netherlands. Consuming chicken was a risk factor for campylobacteriosis caused by chicken-associated STs, whereas consuming beef and pork were protective. Risk factors for campylobacteriosis caused by ruminant-associated STs were contact with animals, barbecuing in non-urban areas, consumption of tripe, and never/seldom chicken consumption. Consuming game and swimming in a domestic swimming pool during springtime were risk factors for campylobacteriosis caused by environment-associated STs. Infections with chicken- and ruminant-associated STs were only partially explained by food-borne transmission; direct contact and environmental pathways were also important. CONCLUSION/SIGNIFICANCE: This is the first case-control study in which risk factors for campylobacteriosis are investigated in relation to the attributed reservoirs based on MLST profiles. Combining epidemiological and source attribution data improved campylobacteriosis risk factor identification and characterization, generated hypotheses, and showed that genotype-based source attribution is epidemiologically sensible.

A quantitative approach towards a better understanding of the dynamics of Salmonella spp. in a pork slaughter-line.

Pork contributes significantly to the public health disease burden caused by Salmonella infections. During the slaughter process pig carcasses can become contaminated with Salmonella. Contamination at the slaughter-line is initiated by pigs carrying Salmonella on their skin or in their faeces. Another contamination route could be resident flora present on the slaughter equipment. To unravel the contribution of these two potential sources of Salmonella a quantitative study was conducted. Process equipment (belly openers and carcass splitters), faeces and carcasses (skin and cutting surfaces) along the slaughter-line were sampled at 11 sampling days spanning a period of 4 months. Most samples taken directly after killing were positive for Salmonella. On 96.6% of the skin samples Salmonella was identified, whereas a lower number of animals tested positive in their rectum (62.5%). The prevalence of Salmonella clearly declined on the carcasses at the re-work station, either on the cut section or on the skin of the carcass or both (35.9%). Throughout the sampling period of the slaughter-line the total number of Salmonella per animal was almost 2 log lower at the re-work station in comparison to directly after slaughter. Seven different serovars were identified during the study with S. Derby (41%) and S. Typhimurium (29%) as the most prominent types. A recurring S. Rissen contamination of one of the carcass splitters indicated the presence of an endemic 'house flora' in the slaughterhouse studied. On many instances several serotypes per individual sample were found. The enumeration of Salmonella and the genotyping data gave unique insight in the dynamics of transmission of this pathogen in a slaughter-line. The data of the presented study support the hypothesis that resident flora on slaughter equipment was a relevant source for contamination of pork.