Spatial scan statistics to assess sampling strategy of antimicrobial resistance monitoring program

The collection and analysis of data on antimicrobial resistance in human and animal populations are important for establishing a baseline of the occurrence of resistance and for determining trends over time. In animals, targeted monitoring with a stratified sampling plan is normally used. However, to our knowledge it has not previously been analyzed whether animals have a random chance of being sampled by these programs, regardless of their spatial distribution. In this study, we used spatial scan statistics, based on a Poisson model, as a tool to evaluate the geographical distribution of animals sampled by the Danish Integrated Antimicrobial Resistance Monitoring and Research Programme (DANMAP), by identifying spatial clusters of samples and detecting areas with significantly high or low sampling rates. These analyses were performed for each year and for the total 5-year study period for all collected and susceptibility tested pig samples in Denmark between 2002 and 2006. For the yearly analysis, both high and low sampling rates areas were significant, with two clusters in 2002 (relative risk [RR]: 2.91, p < 0.01 and RR: 0.06, p < 0.01) and one in 2005 (RR: < 0.01, p < 0.01). For the 5-year analysis, one high sampling rate cluster was detected (RR: 2.56, p = 0.01). These findings allowed subsequent investigation to clarify the source of the sampling clusters. Overall, the detected clusters presented different spatial locations over the years and we can conclude that they were more associated to temporary sampling problems than to a failure in the sampling strategy adopted by the monitoring program. Spatial scan statistics proved to be a useful tool for assessment of the randomness of the sampling distribution, which is important when evaluating the validity of the results obtained by an antimicrobial monitoring program.

Using data on resistance prevalence per sample in the surveillance of antimicrobial resistance

OBJECTIVES

In most existing antimicrobial resistance monitoring programmes, one single bacterial colony from each collected sample is susceptibility tested against a panel of antimicrobials. Detecting the proportion of colonies resistant to different antimicrobials in each sample can provide quantitative data on antimicrobial resistance (resistance prevalence per sample).

METHODS

In this study, a total of 98 faecal samples from slaughter pigs were tested for tetracycline and sulphonamide resistance in Escherichia coli using the single colony method, and these results were compared with the results obtained using the resistance prevalence per sample method.

RESULTS

The results obtained by the resistance prevalence per sample method showed a lower occurrence of resistance. Tetracycline resistance in E. coli was found in 36.7% of the samples using the single colony method, while the mean tetracycline resistance prevalence was 22.5% using the resistance prevalence per sample method. Similarly, sulphonamide resistance was 32.7% using the single colony method and 19.6% when using the resistance prevalence per sample method. Although different estimates were obtained by each method, the correlation test and the regression model demonstrated that there is a significant association between the results obtained using both methods (P value <0.01) for both antimicrobials tested.

CONCLUSIONS

To support risk assessment and analysis of the association between consumption of antimicrobials and occurrence of resistance, there is a need to move towards a more quantitative approach when dealing with antimicrobial resistance in a population, and the resistance prevalence per sample method can provide some of this additional information.

Evaluation of an antimicrobial resistance monitoring program for campylobacter in poultry by simulation

An ideal national resistance monitoring program should deliver a precise estimate of the resistance situation for a given combination of bacteria and antimicrobial at a low cost. To achieve this, decisions need to be made on the number of samples to be collected at each of different possible sampling points. Existing methods of sample size calculation can not be used to solve this problem, because sampling decisions do not only depend on the prevalence of resistance and sensitivity and specificity of resistance testing, but also on the prevalence of the bacteria, and test characteristics of isolation of these bacteria. Our aim was to develop a stochastic simulation model that optimized a national resistance monitoring program, taking multi-stage sampling, imperfect sensitivity and specificity of diagnostic tests, and cost-effectiveness considerations into account. The process of resistance testing of Campylobacter spp. isolated from cloacal swab samples from poultry was modeled using a Markov Chain Monte Carlo model. Different sampling scenarios on the number of flocks to be tested, the number of birds from each flock, and the number of campylobacter colonies submitted to susceptibility testing were evaluated regarding the precision of the resulting prevalence estimate. Precision of the prevalence estimate was defined as the absolute difference between apparent and true prevalence of resistance. A partial budget approach was utilized to find the most cost-effective combination of samples to obtain a defined precision of the prevalence estimate. For a sampling scenario testing 100 flocks, five birds per flock, and one campylobacter colony per sample, the median error of the prevalence estimate was 2.5%, and 95% of the simulations resulted in an error of 7% or less. When the total number of samples was kept constant, maximizing the number of flocks tested, and only testing one bird per flock resulted in the most precise prevalence estimate. Submitting more than one campylobacter colony to resistance testing did not improve the prevalence estimate. Partial budget analysis indicated that the most cost-effective strategy was testing of two birds per flock, and submitting one colony per sample to resistance testing.

A European longitudinal study in Salmonella seronegative- and seropositive-classified finishing pig herds

Surveillance and control are important aspects of food safety assurance strategies at the pre-harvest level of pork production. Prior to implementation of a Salmonella surveillance and control programme, it is important to have knowledge on the dynamics and epidemiology of Salmonella infections in pig herds. For this purpose, 17 finishing pig herds initially classified as seropositive and 15 as seronegative, were followed for a 2-year period through serological and bacteriological sampling. The study included 10 herds from Denmark, 13 from The Netherlands, 4 from Germany and 5 from Sweden and was performed between October 1996 and May 1999. The Salmonella status of finishing pig herds was determined by an initial blood sampling of approximately 50 finishing pigs close to market weight per herd. The development of the Salmonella status of the selected herds was assessed at seven subsequent sampling rounds of 25 blood samples from finishing pigs, 25 blood samples from grower pigs and 10 pen faecal samples each, approximately 3 months apart. The odds for testing finishers seropositive, given that growers were found seropositive previously were 10 times higher than if growers were seronegative (OR 10.0, 95% CI 3.2-32.8). When Salmonella was isolated from pen faecal samples, the herd was more likely to be classified seropositive in the same sampling round, compared to no Salmonella being detected (OR 4.0, 95% CI 1.1-14.6). The stability of an initially allocated Salmonella status was found to vary noticeably with time, apparently irrespective of a seropositive or seronegative classification at onset of the study. Given the measured dynamics in the occurrence of Salmonella in pig herds, regular testing is necessary to enable producers, advisors and authorities to react to sudden increases in the Salmonella prevalence in single herds or at a national level.

Herd-level risk factors for subclinical Salmonella infection in European finishing-pig herds

Our objective was to find herd factors associated with pigs testing seropositive for Salmonella. Data were collected from 359 finishing-pig herds in Germany, Denmark, Greece, The Netherlands and Sweden, between 1996 and 1998. Pigs fed non-pelleted feed (dry or wet) had 2- and 2.5-times lower odds of seropositivity, compared to pigs fed pelleted feed. The protective effect of non-pelleted feed over pelleted feed may be ascribed to the structure and composition. Also, pigs that were given whey (to drink or as the liquid part of the diet) had 2.6-times lower odds to test seropositive than pigs not getting whey. Pigs produced in batches in herds with hygienic-lock facilities had >3-times lower odds for testing seropositive compared to pigs in herds where only one or neither factor was present. In herds where the caretaker(s) washed hands consistently before tending to the animals, pigs had 1.5-times lower odds of seropositivity than pigs in herds where the caretaker did not. Pigs which were able to have snout contact with pigs in neighbouring pens (because pen separations were either open or too low) had 1.7-times higher odds to test seropositive compared to pigs for which such contact was prevented. Pigs in herds recruiting from more than three supplier herds had three-times higher odds to test seropositive than pigs in herds which breed their own replacement stock or recruit from a maximum of three supplier herds.

Recovery of Salmonella enterica from seropositive finishing pig herds

The aim of this study was to assess the probability of detecting Salmonella from pen faecal samples in seropositive classified finishing pig herds. The study involved 77 herds from Denmark (20), The Netherlands (20), Greece (17) and Germany (20). The serological herd status was determined by the blood-sampling of 50 finishing pigs. Bacteriological sampling was performed by 20 pen faecal samples per herd. Over-all, 47% of the blood samples had an OD% larger than 10 and 23% larger than 40. Salmonella was isolated from 135 (9.3%) pen faecal samples in 32 herds (42%). Twenty-eight of these herds (87.5%) had a within-herd seroprevalence larger than 50% at sample cut-off OD% > 10. In our study, there was an increasing probability of recovering Salmonella with increasing within-herd seroprevalence. However, this was only a moderate correlation. A correlation coefficient of 0.62 was found between the proportion of culture positive- and seropositive samples in a herd at cut-off OD%> 10 and of 0.58 at cut-off OD% > 40. Serology is a measure of historical exposure, which may or may not correlate closely to the microbiological burden at the time of sampling. Due to the low sensitivity of culture methods, apparent 'false-positive' serological results may well represent real infections not detected by bacteriological testing. For screening purposes, serological testing provides an indication of exposure to Salmonella, which forms the basis for targeted sampling, intervention and logistic slaughter procedures.