Figures
Abstract
Background
To implement effective control measures, timely outbreak detection is essential. Shigella is the most common cause of bacterial diarrhea in Argentina. Highly resistant clones of Shigella have emerged, and outbreaks have been recognized in closed settings and in whole communities. We hereby report our experience with an evolving, integrated, laboratory-based, near real-time surveillance system operating in six contiguous provinces of Argentina during April 2009 to March 2012.
Methodology
To detect localized shigellosis outbreaks timely, we used the prospective space-time permutation scan statistic algorithm of SaTScan, embedded in WHONET software. Twenty three laboratories sent updated Shigella data on a weekly basis to the National Reference Laboratory. Cluster detection analysis was performed at several taxonomic levels: for all Shigella spp., for serotypes within species and for antimicrobial resistance phenotypes within species. Shigella isolates associated with statistically significant signals (clusters in time/space with recurrence interval ≥365 days) were subtyped by pulsed field gel electrophoresis (PFGE) using PulseNet protocols.
Principal Findings
In three years of active surveillance, our system detected 32 statistically significant events, 26 of them identified before hospital staff was aware of any unexpected increase in the number of Shigella isolates. Twenty-six signals were investigated by PFGE, which confirmed a close relationship among the isolates for 22 events (84.6%). Seven events were investigated epidemiologically, which revealed links among the patients. Seventeen events were found at the resistance profile level. The system detected events of public health importance: infrequent resistance profiles, long-lasting and/or re-emergent clusters and events important for their duration or size, which were reported to local public health authorities.
Author Summary
Shigellosis causes dysentery and kills an estimated 1.1 million people per year worldwide, 60% of them children under the age of 5. The infectious agent is Shigella spp, transmitted from person to person by fecal-oral route or via ingestion of contaminated food or water. Having a system for early detection of outbreaks would be very useful for implementing control measures that help reduce the number of affected patients, economic losses and prevent the dissemination of antimicrobial resistance. We present the application of a space-time permutation scan statistic implemented within the free SaTScan software for laboratory based surveillance of Shigella cases in six provinces from Argentina. SaTScan was applied on the data loaded into WHONET databases (an electronic laboratory data system used world-wide) in the six provinces from April 2009 to March 2012. The project allowed the identification of 32 events, including several of particular public health importance for their duration or number of affected patients. It also strengthened the relationship between the laboratory and epidemiology staff. In conclusion, the combination of WHONET laboratory data and SaTScan analysis can detect important community outbreaks of antimicrobial-resistant shigellosis in a timely manner, to make a difference to public health.
Citation: Viñas MR, Tuduri E, Galar A, Yih K, Pichel M, Stelling J, et al. (2013) Laboratory-Based Prospective Surveillance for Community Outbreaks of Shigella spp. in Argentina. PLoS Negl Trop Dis 7(12): e2521. https://doi.org/10.1371/journal.pntd.0002521
Editor: Margaret Kosek, The Johns Hopkins University, United States of America
Received: January 15, 2013; Accepted: September 6, 2013; Published: December 12, 2013
Copyright: © 2013 Viñas et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: Support was provided by cooperative agreement U01GM076672 from the National Institutes of Health under the Models of Infectious Disease Agent Study (MIDAS) programme and research project grant RR025040 from the National Institutes of Health. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
In view of the increasing movement of people, animals, and food products around the globe, new strategies and collaborations are urgently needed to detect the emergence of microbial threats and implement effective control measures. National and regional electronic laboratory-based surveillance collaborations based on routine clinical laboratory test results, as recommended by the WHO-Global Foodborne Infections Network [1] and the WHO-Advisory Group on Integrated Surveillance of Antimicrobial Resistance [2], offer the potential for real-time monitoring of evolving microbial populations. Sophisticated technologies for differentiating among strains and for processing information have proliferated and been incorporated into surveillance [3]. However, advances in organizational aspects such as timeliness of data entry and analysis and integration of local site-of-care laboratories into national and international surveillance networks have developed more slowly [4]. Statistical analysis of laboratory data for the detection of disease outbreaks in the community or in hospitals has also lagged, and existing statistical approaches have tended to focus on temporal trends [5], [6], [7], [8], largely ignoring the geographic component of pathogen population dynamics.
To be most useful for public health purposes, laboratory-based surveillance should 1) be specific, i.e. be capable of distinguishing (a) among species and preferably variants within species and (b) among antimicrobial resistance profiles within those taxonomic groups; 2) have timely electronic data entry; 3) integrate multiple laboratories using uniform protocols and a uniform database; 4) be linked to and used by the agencies responsible for disease control; and 5) implement statistical methods for detecting departures from background levels in both time and space, rather than relying solely on visual inspection of data.
Of the bacterial pathogens causing diarrhea, Shigella spp. is one of the most prevalent and most consistently associated with dysentery and persistent diarrhea [9]. Shigellosis kills an estimated 1.1 million people per year worldwide, 60% of them children under the age of 5 [10], and can result in reduced growth in children who survive. Shigella species appear highly adaptable to selective pressure and have developed resistance to a number of antimicrobials with patterns of resistance varying temporally and geographically with antimicrobial usage patterns [11], [12], [13], [14]. Highly resistant clones of Shigella have emerged in Argentina [15], [16], [17]. Recently, the unique Shigella flexneri serotype X variant, which emerged in China in 2001, has rapidly spread, including through Argentina [18], undergoing frequent serotype switching and acquiring resistance to multiple antimicrobials in the process [19].
We report here our experience with an evolving, integrated, laboratory-based, near real-time surveillance system now operating in six contiguous provinces of Argentina, building on a prior retrospective study [20]. It represents real-world surveillance rather than a static system pre-defined in a formal protocol. This system includes all of the desired laboratory-based surveillance system elements listed above and may serve as a model for surveillance not only of Shigella spp. but of other community-acquired pathogens in Argentina and elsewhere.
Materials and Methods
Surveillance population and period
The microbiology data used for shigellosis prospective surveillance came from a subset of the national Argentine network for monitoring antimicrobial resistance, WHONET- Argentina, which was established in 1986 by the Ministry of Health through the Dr. Carlos G. Malbrán National Institute of Infectious Diseases (INEI). The group is named after WHONET, a free software for the management of microbiology laboratory data developed by the WHO Collaborating Centre for Surveillance of Antimicrobial Resistance promoted by the World Health Organization. Led by the National Reference Laboratory at INEI, WHONET-Argentina currently includes 89 clinical laboratories representing all geographic jurisdictions and captures detailed data on human pathogens and their susceptibility profiles from routine diagnostic specimens.
WHONET-Argentina hospitals were selected to participate in this surveillance initiative on the basis of the completeness and timeliness of data entered into the national WHONET database in previous years, a commitment to send updated data on a weekly basis, and Shigella serotyping ability. An additional criterion was the strength of the respective local public health systems in integrating laboratory, epidemiology, food science, and environmental health efforts for the investigation of outbreaks and sources of infection.
On the other hand, surveillance of Shigella infections at the national level is currently conducted by provincial hospitals that report the number of cases weekly to the Ministry of Health.
The period covered by this report was from April 1, 2009 through March 31, 2012. The number of laboratories and provinces participating increased during this period (Table 1), with growth in laboratories' historical data, capacity for complete and timely data entry, and ability to meet the other inclusion criteria. For the first year, seven hospitals in the three contiguous provinces of La Pampa (LP), Neuquén (NQ), and Río Negro (RN) participated. In March 2010, four satellite clinics in the same three provinces were added. In January 2012, 12 additional laboratories from the three additional contiguous provinces of Córdoba (CBA), Mendoza (MZA), and San Luís (SL) were incorporated (Table 1). Thus, by the end of the evaluation period, 23 laboratories in six provinces were included (Figure 1).
The 23 participating laboratories of six provinces of Argentina are indicated with a green circle : 11 hospitals in the three contiguous provinces HAC, BAR, HAZ and HGR from Río Negro province (RN); HLM, HGC from La Pampa province (LP); HHH, NEU, HJA, HNB, SMA from Neuquén province (NEU); and with a red circle: 12 additional hospitals from the three additional contiguous provinces: PJC, RFB, IMC, HST, VIL, HDF and VSW from Córdoba province; PVM and HSL of San Luis province and H09, NOT and HTS from Mendoza province.
Microbiological characterization and data-processing
Data collection at the participating laboratories.
Stool specimens taken from patients with diarrhea seeking healthcare at participating hospitals and clinics were investigated for the presence of enteric pathogens. Isolation and biochemical identification of Shigella spp. were performed using conventional methods [21]. All strains were serotyped using antisera provided by the National Institute of Biological Production (INPB), part of the National Administration of Health Laboratories and Institutes (ANLIS). Susceptibility tests were performed by the disk diffusion method according to Clinical and Laboratory Standards Institute recommendations of 2012 [22]. Susceptibility to the following drugs was evaluated: ampicillin, trimethoprim/sulfamethoxazole, nitrofurantoin, ciprofloxacin, nalidixic acid, cefpodoxime, and fosfomycin, as specified in the annually reviewed Working Protocol of WHONET - Argentina [23].
Microbiological and demographic details of the shigellosis cases were entered into the WHONET 5.6 microbiology database at each participating laboratory. These included patient date of birth, sex, clinical diagnosis, epidemiologic risk factors, co-morbidities, specimen date and type, the bacterial species isolated, serotype, and resistance phenotype for the above-mentioned antimicrobials. A WHONET filter was applied to include only the first sample from each patient in a 60-day period. The latitude and longitude of each participating laboratory and clinic were obtained from Google Earth using the postal address.
Data analysis at the central laboratory.
Each participating hospital sent its full WHONET database to the National Reference Laboratory at INEI on the Wednesday following the end of each Sunday-to-Saturday epidemiologic week. At INEI, these data were analyzed every Thursday to look for statistical signals suggesting clustering of shigellosis cases in time and space as described in the statistical section below. Signal detection analyses were performed at several taxonomic or genotypic levels: for all Shigella spp. aggregated together, for each of the individual species (S. flexneri, S. sonnei, S. boydii, and S. dysenteriae), for serotypes within species (S. flexneri serotype 1, 2, etc.), and for antimicrobial resistance phenotypes within species. Analyses at higher taxonomic levels can have greater statistical power to find clusters at more specific levels due to incorporating more information about background rates. Even though susceptibility to seven antimicrobials was routinely tested by clinical laboratories, most Shigella isolates were susceptible to all agents, with two exceptions: ampicillin (AMP) and trimethoprim/sulfamethoxazole (SXT). Therefore, at the beginning of the collaboration, the focus of analyses was on the four possible phenotypes with these two agents: susceptible to both, resistant to both, resistant to AMP but not SXT, and resistant to SXT but not AMP. In response to the recognition by WHONET-Argentina coordinators of growing resistance to cefpodoxime (CPD), analysis of the eight resistance phenotype combinations with all three agents (AMP, SXT, and CPD) was initiated in June 2011 (Table 1).
During the period of this study, a new serotype of S. flexneri was identified in Argentina, recently recognized as a S. flexneri serotype X variant [Viñas M, van der-Ploeg C et al., unpublished data]. Initially, participating laboratories entered these isolates into WHONET as atypical Shigella flexneri with “fxx” code indicating non-typeable S. flexneri. A new code “479” (referring to the antiserum developed by the INPB) was created to permit specific tracking of this new serotype named S. flexneri AA479 (Table 1). This code has been used systematically by participating laboratories since January 2012.
Molecular strain typing.
Shigella isolates associated with statistically significant signals identified with the cluster detection method described below were sent from hospital laboratories to INEI and subtyped by pulsed field gel electrophoresis (PFGE) for comparison with each other and with other isolates recorded in the national database (NDB). The PulseNet standardized protocols for S. sonnei [24] and S. flexneri [25] were applied. The NDB was created by the application of these protocols with the use of a universal standard (S. Braenderup) for normatization of the gels in the frame of the PulseNet network. Briefly, genomic DNA embedded in agarose plugs was digested with XbaI enzyme for S. sonnei and with NotI enzyme for S. flexneri, and generated fragments were separated by electrophoresis with linear switch times of 2.54 to 54 sec. and 5 to 35 sec for S. sonnei and S. flexneri, respectively. Genetic relatedness between the isolates that showed indistinguishable PFGE profiles with those enzymes was confirmed with a second enzyme, BlnI for S. sonnei and XbaI for S. flexneri. The results were analyzed using BioNumerics software (Applied Maths), and dendrograms were constructed applying the DICE coefficient and UPGMA (Unweighted Pair Group Method with Arithmetic Mean), with a band position tolerance window of 1.5% as established by PulseNet. The criteria for interpretation of PFGE considered the variability of the species under study, the epidemiological context and the frequency of the patterns identified based on information in the national database [26]; also we applied the criteria by Tenover et al. that allowed to assign the isolates to categories of relatedness based on the difference in the number of bands [27].
Space-time permutation scan statistic
To detect localized shigellosis outbreaks in near real-time, we used the non-parametric space-time permutation scan statistic [28], as was previously used in a pilot study of historical data in the WHONET-Argentina database . The method searches for statistically significant clusters of Shigella cases in space and time, using cylindrical scanning windows with a circular base of variable location and radius representing geographical space and a variable height representing the number of days in a potential cluster (ending on the day for which the analysis is being done). It does not require population-at-risk data and makes minimal assumptions about the time, location, or size of outbreaks. The method adjusts for any purely geographical variation in disease incidence, whether due to urban vs. rural, north vs. south, dry vs. humid conditions, etc. There is no need to adjust for any of these variables explicitly. Rather, the adjustment is done non-parametrically, using a permutation-based approach that is conditional on the total area counts summed over all days. Similarly, the method adjusts for any purely temporal trends in the data, such as seasonal variation or day-of-week effects. This is done non-parametrically, by day, by using a permutation-based approach that is conditional on the total daily counts summed over all geographic areas. With a seasonal infection such as shigellosis, this ability to automatically adjust for seasonal variation is critical; without such an adjustment it would be difficult to distinguish between epidemiologically important outbreaks and the expected annual increase in shigellosis cases observed each summer. In contrast to other cluster detection methods that adjust for seasonal variation using a parametric model, the space-time permutation scan statistic does not require multiple years of historical data. The paper by Kulldorff et al. [28] describes the method in more detail.
In looking for clusters, we set the maximum temporal cluster length at 30 days, meaning clusters of 1, 2, 3, and up to 30 days' duration could be detected. Because data were updated on a weekly rather than daily basis and because of potential delays in data entry, data transmission, or data availability (e.g. in the ascertainment of organism serotype), prospective analyses were run not only for the last date in the dataset but also for each day in the prior several weeks to ensure that recent clusters were not missed. In each of these day-specific analyses, the prior 365 days of data were used as a historical baseline. From April 2009 to August 2011 we used SaTScan 7.0.1 to search for single hospital clusters; from September 2011 through March 2012, we used SaTScan 9.1.1 to search for single or multi-hospital clusters with a maximum radius of 200 kilometers (Table 1) (Clusters of that apparent or real size could occur if a contaminated food were distributed regionally or if an area of transmission were located between two participating hospitals, leading some patients to go to one hospital and others to the other).
Per analysis run, the statistical inference is adjusted for the multiple testing inherent in the many potential cluster locations, sizes, and time lengths, and is expressed in terms of a recurrence interval [29]. If a detected group of cases is determined to have a recurrence interval of 400 days, for example, this could be an event of epidemiologic significance. But it could be simply due to chance -during any 400-day period, the expected number of signals of that strength or stronger is one, when the null hypothesis of no clusters is true. Thus, the higher the recurrence interval, the less likely that the observed clustering could be attributed to random variation. In this study, we considered a grouping of cases with a recurrence interval of 365 days or longer to be a statistical “signal” and worth communicating (after basic data quality checking) to relevant local or provincial authorities for possible epidemiologic and molecular investigation. Using 365 days as our recurrence interval threshold for notifying public health authorities means that, on average, we could have expected to see one false positive signal per year within each analysis level (genus, species, serotype, and resistance profile). Since statistically significant space-time groupings of cases may be caused by chance or by organizational or procedural factors such as changes in hospital participation, specimen collection practices, or laboratory testing procedures, it is important that statistical signals be evaluated through traditional epidemiologic means before concluding that they are indications of true disease outbreaks.
Calculations for the space-time permutation scan statistic were done using the free SaTScan™ software [30], as imbedded in the free WHONET software.
Terminology
In the rest of this paper, we use the term “signal” to refer to the detection by SaTScan of a group of cases clustered in space and time with a recurrence interval of 365 days or more. We use the term “event” to refer to a group of signals overlapping in space and time, which may represent a single potential disease outbreak. The term “cluster” is used generically, referring to a group of cases regardless of whether or not these cases are truly related to each other epidemiologically.
Epidemiologic investigation
When a signal was detected in weekly analyses, a report including data on each patient was sent by INEI analysts to laboratory and epidemiology personnel in the affected hospital and province. These messages were usually sent on Fridays, but later if there were delays in analysis.
The decision by local authorities of whether to investigate a signal depended in part on the recurrence interval and in part on other criteria such as number of patients, location details, and timing and specificity of the signal (homogeneous clusters detected by resistance phenotype or serotype were considered to be more reliable than heterogeneous clusters detected at the Shigella genus level). Since the purpose of this real-world surveillance system was to support local public health, no uniform signal investigation protocol was imposed; investigations variously included patient or clinician surveys, additional sampling of either contacts or suspected sources (food or water) for Shigella spp., and trace-back of suspected food products.
The events identified by SaTScan confirmed as outbreaks with investigation by local authorities and the confirmation of strains' relatedness by PFGE were analyzed and informed to National Surveillance System.
Results
There were 32 statistically significant events: 2, 12, 3 and 2 of S. flexneri serotypes 1, 2, 3, and AA 479 respectively; 11 of S. sonnei; and 2 of S. boydii serotype 2 (Table 2 and Table 3). Seventeen of the 32 events were found at the resistance profile level (and some at higher levels also). All 6 participating provinces had events. During the study period, 21 of the events occurred primarily during the Argentine summer months of October–March, while 11 occurred primarily in April–September. The median event duration was 48.5 days (minimum: 3, maximum: 94). The median number of patients in a signal was 21.5 (minimum: 2, maximum: 41).
Twenty-six of the 32 events were investigated further by PFGE analysis, which confirmed a close relationship among the isolates (with the first and second enzyme) for 22 (84.6%) of the 26. In contrast, in Events 10, 11, 12, and 28 (all involving S. sonnei, PFGE showed diverse genetic subtypes, and we considered these events to be largely chance concentrations of cases, not potential diseases outbreaks. Seven of the events, Events 5, 7, 14, 16, 17, 31, and 32, were investigated epidemiologically, which revealed links among the patients (Table 4), consistent with the PFGE findings for these events.
Twenty-six of the 28 events considered to represent or possibly represent disease outbreaks were detected before hospital staff was aware of any increase in the number of Shigella isolates. The others were Events 14 and 17. In Event 14, S. flexneri 2 in Rio Negro, the hospital bacteriologist suspected the outbreak before the appearance of the first WHONET- SaTScan signal, because the majority of isolates were from patients from the same family. Event 17 was a cluster of S. flexneri 3 associated with a wedding in Neuquén province, which was attended by 150 people, including some from Chile. More than half the participants became ill, and this outbreak was reported immediately to public health authorities at both national and international levels before the laboratory results could be incorporated into the WHONET database and analyzed.
The system detected events of public health importance. For example, Events 16 and 21, both in Rio Negro, involved S. sonnei resistant to both AMP and SXT, an infrequent resistance profile. A number of long-lasting and/or re-emergent clusters were also detected, represented by 4 pairs of related events: PFGE patterns were very similar for Events 1 and 6, S. flexneri 2 in La Pampa (Figure 2), and for Events 13 and 14, S. flexneri 2 in Rio Negro. Events in each pair were separated from each other by at least 3 months. Events 10 and 11 of S. sonnei in La Pampa were separated by one month and included two persistent patterns and single subtypes circulating simultaneously. Events 16 and 21, S. sonnei with the unusual AMP-SXT resistance phenotype, were centered in two cities (Viedma and Bariloche) in Rio Negro province. The predominant PFGE patterns in the latter two events differed by only one band, which is not considered a significant difference in PFGE analysis for Shigella when person-to-person transmission is a prominent feature and the outbreak persists in time [26] Event 16 lasted from December 2010 to March 2011, while Event 21 occurred in April to May 2011.
Isolates recovered in Santa Rosa, La Pampa (Event 1 and 6), and selected isolates for comparison, including one from an outbreak in San Luis and three from sporadic cases in 2009. The rectangle highlights isolates of S. flexneri 2 recovered in March–April 2009 (Event 1) and October–November 2009 (Event 6).
Two other events are worthy of mention due to their duration or size. Event 7, caused by SXT-nonsusceptible S. sonnei in La Pampa province, lasted from December 2009 to March 2010, with 34 isolates. Among 23 cases of diarrhoea studied trough clinician surveys, 14 (60.9%) were epidemiologically linked. Seven of 14 SXT-non-susceptible S. sonnei isolates analyzed by PFGE shared a new pattern in the national database and the other 7 were closely related to this pattern with similarities between 91.4% to 97.4%, 1 to 3 bands of difference. (Figure 3). No common source was identified, but most of the cases were associated with two neighboring households found to be epidemiologically linked and to have deficient sanitary conditions. Figure 4 shows the time series of cases, for the period from December 2009 to March 2012, to highlight the detection and evolution of event 7. Event 17, the wedding outbreak in Neuquen province, was noteworthy in that it was detected by our surveillance system on the basis of only 7 isolates when in fact the total number ill, according to public health investigation, was closer to 75. The 7 S. flexneri 3 isolates from the patients showed indistinguishable PFGE patterns.
Isolates recovered in Santa Rosa, La Pampa in January – February 2010. The rectangle highlights the 7 S. sonnei isolates with an indistinguishable PFGE pattern which had not previously been recorded in the National Data Base (NDB). The remaining 7 isolates identified statistically and epidemiologically as part of Event 7, exhibited high genetic relatedness (from 91.4 to 97.4% similarity, 1 to 3 bands of difference) to the most frequent pattern within the event, confirming the relation of the cases.
The rectangle highlights the time series of cases corresponding to the detection and evolution of Event 7.
In general, no common source could be confirmed in the events, even though food and water samples were analyzed in several instances; this may be due to the difficulty for the isolation of Shigella from this kind of sample. Nevertheless, the epidemiologic studies could determine sanitary deficient conditions and probable routes of transmission, mainly from person to person. Furthermore, deficiencies in the conditions for food conservation and elaboration were identified in some events. On this basis, control and prevention measures included recommendations on hygiene and food handling, as well as a notification to the International Health Regulation for an event that affected patients from Chile and Argentina (Table 4).
Discussion
In three years of active, near real-time surveillance, building on an earlier, purely retrospective pilot study [28], our system detected 32 shigellosis events ( Table 2 and 3 ). Independent suspicion or discovery of only 2 of the 28 events considered suspect of outbreak occurred prior to detection of the first signal by WHONET-SaTScan. Of the 26 events for which we have PFGE evidence, 22 appeared to represent groups of truly genetically interrelated cases, including 9 new patterns (1 of them closely related to pattern identified previously) and 13 subtypes identified before in the NDB , with supporting evidence of epidemiologic linkage for 7.
The 32 detected events represented a broad range of the Shigella variants circulating in Argentina and were distributed among the six participating provinces. Some were of particular public health importance because of long duration or number of patients, e.g. Events 7 and 16, or because of a distinct resistance profile, e.g. Events 16 and 21, both S. sonnei AMP-SXT. Four pairs of events that were related according to PFGE patterns may have represented additional long-lasting outbreaks.
The two known outbreaks in these six provinces that were not detected by WHONET-SaTScan could not have been found by the system. One, a plasmid-conferred cefpodoxime-resistant cluster of S. sonnei in February-March 2011, could not have been detected because cefpodoxime resistance was not one of the phenotypes analyzed at the time. However, it did show up as a non-statistically significant cluster of SXT-AMP-nonsusceptible resistant S. sonnei, with RI = 116. The other, of Shigella flexneri AA479, appeared in one of the new provinces in August–September 2011, but this was before the new laboratories' data were incorporated and analyzed and before the variant was given a specific code in WHONET.
In Argentina, under the auspices of the international laboratory network PulseNet, it has been possible to maintain active surveillance using PFGE to detect circulating clones [16], [31]. When PFGE results are communicated to local public health agencies, they inform investigation into possible sources of contamination and their persistence over time [32]. PFGE results of the type we often saw one predominant pattern with other closely related genetic subtypes in the same event are common in Shigella spp., particularly in long-lasting events with person-to-person transmission, such as Events 7 and 16. Others have reported widespread outbreaks in which this mode of transmission was confirmed by molecular typing results and epidemiologic data [33], [34]. Where person-to-person spread is a prominent feature of the outbreak, more variability is expected [26], [27]. In each of two outbreaks with point-source exposures, event 17 and the outbreak of S. flexneri AA479 in the provinces that had not yet been incorporated into the surveillance system, the PFGE patterns were indistinguishable.
There are several limitations to this study. As is often the case in evaluations of surveillance systems, there was no known set of outbreaks that could serve as a gold standard against which to compare all the events. Some events were studied by public health local authorities, while all the cases were reported to the national surveillance system. Therefore, it was not possible to calculate such measures as sensitivity, specificity, or negative predictive value. PFGE testing did stand in as a strong validation method, and 22 (84.6%) of 26 events for which PFGE was done showed evidence of close genetic relatedness. However, close relatedness by itself does not prove that isolates belong to a single outbreak, and accompanying epidemiologic studies were not carried out for every event, so we cannot claim to know the positive predictive value of our system. Also, the system was not pre-specified in a static protocol, having changed in important ways over the three years: the inclusion of additional laboratories, changes in parameter settings, and changes in the kinds/phenotypes of Shigella for which clusters could be detected (Table 1). The results may have looked somewhat different had we included all 23 laboratories and the final parameter settings and Shigella variants from the start. Finally, we did not compare results of different cluster detection methods applied to the same data. We selected the space-time permutation test because it does not require population-at-risk (denominator) data and it adjusts for purely spatial and purely temporal variation and can do so without multiple years of historical data. It would be worthwhile in a future methods-oriented endeavour to compare the performance of this method with other cluster detection methods.
In large measure, this still-evolving, real-world, laboratory-based surveillance system satisfies criteria for public health utility, including that it 1) be specific, 2) have timely electronic data entry, 3) integrate multiple laboratories using uniform protocols and databases, 4) be used by the agencies responsible for disease control, and 5) implement statistical methods for detecting departures from background levels in both time and space. We have detected clusters of shigellosis of public health importance, which have been confirmed by PFGE as consisting of closely related clones, and informed local public health efforts. This WHONET-SaTScan system of data organization and analysis could represent a good complementary tool for national surveillance system, for early outbreak detection in real time, signalling the importance to investigate some events , and could be applied to other pathogens, implemented by other networks of laboratories, and scaled up to national and international levels.
Acknowledgments
We thank Raquel Terragno for providing expert advice and for encouraging this study in the country and collaboration of Group MIDAS – Argentina.
Members of Group MIDAS - Argentina
Working Group MIDAS (Modeling Infectious Disease Agent Study) – Argentina.
Neuquén: H. Sauer, Hospital ”Dr H. Heller” (HHH); A. Zurschmitten, Hospital Junín de los Andes (HJA); F. Hadad, Epidemiología del HJA, G. González, M.R. Nuñez, Hospital Provincial de Neuquén “Dr Castro Rendón” (NEU); N. Guyot, Hospital Centenario “Dr Natalio Burd” (HNB); J. Rojas, Hospital San Martín de Los Andes (SMA); C. López, Servicio de Epidemiología del Hospital Provincial de Neuquén “Dr Castro Rendón” (NEU); L. Pianciola, Laboratorio Central, Subsecretaría de Salud de Neuquén; G. Gottardi, Dirección de Bromatología de Neuquén; E. Oller, Coordinación Red Lab Neuquén.
Río Negro: C. Carranza, Hospital A. Cipolletti (HAC); D. Durany, Hospital General Roca (HGR); G. Rivollier, G. Stafforini, Hospital A. Zatti (HAZ); L. Acerbi, Epidemiología del HAZ; G.S. De Bunder, Hospital Zonal de Bariloche (BAR); L. Iglesias, M. López, Saneamiento Ambiental, Bromatología Bariloche; M. Nóbile, Coordinación Red Lab Río Negro.
La Pampa: G. Almada, Hospital Lucio Molas (HLM); A. Pereyra, I. Sylveira, Hospital Gobernador Centeno (HGC); L. Ottavianoni, Bromatología La Pampa; C. Rechimont, Epidemiología La Pampa; C. Elorza, Coordinación Red Lab La Pampa.
Mendoza: M.A. Distefano, Hospital Central de Mendoza (H09); B. García, Hospital Notti (NOT); C. Baldoni, Hospital Schestakow (HTS); H. Pagella, Coordinación Red Lab Mendoza.
San Luis: E. Fernández de Vargas, Hospital Regional de Villa Mercedes (PVM); H. Rigo, Policlínico Central de San Luis (HSL); A. Giunta, Epidemiología. Coordinación Red Lab San Luis.
Córdoba: M.E. Suárez, Lab. Central de Córdoba; A. Rondini, J. Casado, Bromatología; L. González, Hospital Infantil Municipal de Córdoba (IMC); C. Culasso, Hospital de Niños de la Santísima Trinidad (HST); P. Cortés, Hospital Pediátrico del Niño Jesús (PJC); M. Botiglieri, Clínica Privada Reina Fabiola (RFB); C. Aimaretto de Costabella, Hospital Villa María (VIL); L. Camisassa, Hospital Regional Domingo Funes (HDF); L. Wolff, Clínica Vélez Sarsfield (VSW); P. Esquivel, Epidemiología Córdoba; A. Cudolá, Coordinación Red Lab Córdoba.
Author Contributions
Conceived and designed the experiments: JS MK MG. Performed the experiments: ET AG MG MRV SPB ADG MP. Analyzed the data: MRV ET AG KY MP SPB ADG JS MK MG. Contributed reagents/materials/analysis tools: SPB ADG CvdP SB AR MIC. Wrote the paper: MRV KY JS MK MP MG.
References
- 1.
WHO Global Foodborne Infections Network (GFN). Data and statistics. Available at: http://www.who.int/gfn/en/. Accessed 8 October 2012.
- 2.
WHO Advisory Group on Integrated Surveillance of Antimicrobial Resistance (AGISAR). Antimicrobial Resistance Available at:http://www.who.int/foodborne_disease/resistance/agisar/en/. Accessed 8 October 2012
- 3. Swaminathan B, Gerner-Smidt P, Ng LK, Lukinmaa S, Kam KM, et al. (2006) Building PulseNet International: an interconnected system of laboratory networks to facilitate timely public health recognition and response to foodborne disease outbreaks and emerging foodborne diseases. Foodborne Pathog Dis Spring 3 (1) 36–50.
- 4. O'Brien TF, Stelling J (2011) Integrated multilevel surveillance of the world's infecting microbes and their resistance to antimicrobial agents. Clin Microbiol Rev Apr; 24 (2) 281–95
- 5. Hutwagner LC, Maloney EK, Bean NH, Slutsker L, Martin SM (1997) Using laboratory-based surveillance data for prevention: an algorithm for detecting Salmonella outbreaks. Emerg Infect Dis 3 (3) 395–400.
- 6. Widdowson MA, Bosman A, van Straten E, Tinga M, Chaves S, et al. (2003) Automated, laboratory-based system using the Internet for disease outbreak detection, the Netherlands. Emerg Infect Dis 9 (9) 1046–52.
- 7. Brown SM, Benneyan JC, Theobald DA, Sands K, Hahn MT, et al. (2002) Binary cumulative sums and moving averages in nosocomial infection cluster detection. Emerg Infect Dis 8 (12) 1426–32.
- 8. Hacek DM, Cordell RL, Noskin GA, Peterson LR (2004) Computer-assisted surveillance for detecting clonal outbreaks of nosocomial infection. J Clin Microbiol 42 (3) 1170–5.
- 9. Kosek M, Peñataro Yori P, Paredes Olortegui M (2010) Shigellosis update: advancing antibiotic resistance, investment empowered vaccine development, and green bananas. Current Opinion in Infectious Diseases 23: 475–480.
- 10. Kotloff KL, Winickoff JP, Ivanoff B, Clemens JD, Swerdlow DL, et al. (1999) Global burden of Shigella infection: Implications for vaccine development and implementation of control strategies. Bull. World Health Organ 77: 651–666.
- 11. Vinh H, Nhu NT, Nga TV, Duy PT, Campbell JI, et al. (2009) A changing picture of shigellosis in southern Vietnam: shifting species dominance, antimicrobial susceptibility and clinical presentation. BMC Infect Dis Dec; 9: 204.
- 12. Khatun F, Faruque AS, Koeck JL, Olliaro P, Millet P, et al. (2011) Changing species distribution and antimicrobial susceptibility pattern of Shigella over a 29-year period (1980–2008). Epidemiol Infect 139 (3) 446–452.
- 13. Rosewell A, Ropa B, Posanai E, Dutta SR, Mola G, et al. Shigella spp. antimicrobial drug resistance, Papua New Guinea, 2000–2009 (2010). Emerg Infect Dis 16 (11) 1797–1799
- 14. Vlieghe E, Phoba MF, Tamfun JJ, Jacobs J (2009) Antibiotic resistance among bacterial pathogens in Central Africa: a review of the published literature between 1955 and 2008. Int J Antimicrob Agents Oct; 34 (4) 295–303 Epub 2009 Jun 21.
- 15. Andres P, Petroni A, Faccone D, Pasterán F, Melano R, et al. (2005) Extended-spectrum beta-lactamases in Shigella flexneri from Argentina: first report of TOHO-1 outside Japan. Int J Antimicrob Agents; Jun 25 (6) 501–7.
- 16. Pichel M, Gonzalez Fraga S, Terragno R, Mulki J, Gentile A, et al. (2007) Short report: analysis of clonal relationship among Shigella sonnei isolates circulating in Argentina. Epidemiol Infect 135 (4) 681–7.
- 17. Rapoport M, Monzani V, Pasteran F, Morvay L, Faccone D, et al. (2008) CMY-2-type plasmid-mediated AmpC beta-lactamase finally emerging in Argentina. Int J Antimicrob Agents 31 (4) 385–7.
- 18.
Viñas MR, Brengi SP, Van der Ploeg CA, Caffer MI, Bordagorria XL, et al.. (2012) Caracterización feno-genotípica de aislamientos de Shigella flexneri atípicos identificadas por el sistema de vigilancia de diarreas en Argentina. In: VII Congreso de la Sociedad Argentina de Bacteriología, Micología y Parasitología Clínicas - SADEBAC, 01 (74).
- 19. Ye C, Lan R, Xia S, Zhang J, Sun Q, et al. (2010) Emergence of a new multidrug-resistant serotype X variant in an epidemic clone of Shigella flexneri. J Clin Microbiol Feb; 48 (2) 419–26.
- 20. Stelling J, Yih WK, Galas M, Kulldorff M, Pichel M, et al. (2010) Automated use of WHONET and SaTScan to detect outbreaks of Shigella spp. using antimicrobial resistance phenotypes. Epidemiol Infect 138 (6) 873–883
- 21.
Ewing WH (1986) Chapter 7. The genus Shigella. In: Edwards W and Ewing's P Identification of Enterobacteriaceae, 4th Edition. New York: Elsevier Science Publishing Co., Inc. pp.135–172.
- 22.
Clinical and Laboratory Standards Institute (CLSI) (2012) Performance Standards for Antimicrobial Susceptibility Testing; Twenty-Second Informational Supplement. M100-S22. Wayne, PA.
- 23.
WHONET, Protocolo de trabajo XIII Taller WHONET - Argentina (2010). Available at: htpp:// www.whonet.org/DNN/LinkClick.
- 24. Ribot EM, Fair MA, Gautom R, Cameron DN, Hunter SB, et al. (2006) Standardization of pulsed-field gelelectrophoresis protocols for the subtyping of Escherichia coli O157:H7, Salmonella, and Shigella for PulseNet. Foodborne Pathog Dis 3: 59–67.
- 25. Pichel M, Brengi S, Cooper KLF, Ribot EM, Al-Busaidy S, et al. (2012) for the Shigella flexneri PulseNet PFGE Protocol Working Group (2012) Standardization and International Multicenter Validation of a PulseNet Pulsed-Field Gel Electrophoresis Protocol for Subtyping Shigella flexneri. Isolates Foodborne Pathogens and Disease 9 (5) DOI:https://doi.org/ 10.1089/fpd.2011.1067.
- 26. Barrett TJ, Gerner-Smidt P, Swaminathan B (2006) Interpretation of pulsed-field gel electrophoresis patterns in foodborne disease investigations and surveillance. Foodborne Pathog Dis Spring 3 (1) 20–31.
- 27. Tenover FC, Arbeit RD, Goering RV, Mickelsen PA, Murray BE, et al. Interpreting chromosomal DNA restriction patterns produced by pulsed-field gel electrophoresis: criteria for bacterial strain typing. J Clin Microbiol 1995 September; 33 (9) 2233–2239.
- 28. Kulldorff M, Heffernan R, Hartman J, Assunção RM, Mostashari F (2005) A space-time permutation scan statistic for the early detection of disease outbreaks. PLoS Medicine 2: 216–224.
- 29. Kleinman K, Lazarus R, Platt R (2004) A generalized linear mixed models approach for detecting incident clusters of disease in small areas, with an application to biological terrorism. Am J Epidemiol 159: 217–24.
- 30.
Kulldorff M (2011) SaTScan,version 9.1.1. Available at: http://www.satscan.org. Accessed 8 October 2013.
- 31.
Brengi SP, Della Gaspera A, Viñas MR, Panagópulo M, Caffer MI, et al..(2012) Epidemiología molecular aplicada a la vigilancia: base de datos nacional de electroforesis en campo pulsado de Shigella sonnei y Shigella flexneri. In: VII Congreso de la Sociedad Argentina de Bacteriología, Micología y Parasitología Clínicas (SADEBAC), O5 (82)
- 32. Lee YS, Liu MC, Ko CF, Lu CH, Tseng YH (2005) Molecular epidemiology of Shigella flexneri in a long-stay psychiatric nursing center during 2001 to 2003. J Clin Microbiol Mar; 43 (3) 1353–60.
- 33. Chiou CS, Hsu WB, Wei HL, Chen JH (2001) Molecular epidemiology of a Shigella flexneri outbreak in a mountainous township in Taiwan, Republic of China. J Clin Microbiol Mar; 39 (3) 1048–56.
- 34. Chen JH, Chiou CS, Chen PC, Liao TL, Liao TL, et al. (2003) Molecular epidemiology of Shigella in a Taiwan township during 1996 to 2000. J Clin Microbiol Jul; 41 (7) 3078–88.