Mining free-text medical records for companion animal enteric syndrome surveillance

Text mining for disease surveillance in veterinary clinical data: part one, the language of veterinary clinical records and searching for words

The development of natural language processing techniques for deriving useful information from unstructured clinical narratives is a fast-paced and rapidly evolving area of machine learning research. Large volumes of veterinary clinical narratives now exist curated by projects such as the Small Animal Veterinary Surveillance Network (SAVSNET) and VetCompass, and the application of such techniques to these datasets is already (and will continue to) improve our understanding of disease and disease patterns within veterinary medicine. In part one of this two part article series, we discuss the importance of understanding the lexical structure of clinical records and discuss the use of basic tools for filtering records based on key words and more complex rule based pattern matching approaches. We discuss the strengths and weaknesses of these approaches highlighting the on-going potential value in using these "traditional" approaches but ultimately recognizing that these approaches constrain how effectively information retrieval can be automated. This sets the scene for the introduction of machine-learning methodologies and the plethora of opportunities for automation of information extraction these present which is discussed in part two of the series.

Can North American animal poison control center call data provide early warning of outbreaks associated with contaminated pet food? Using the 2007 melamine pet food contamination incident as a case study

The 2007 melamine pet food contamination incident highlighted the need for enhanced reporting of toxicological exposures and development of a national quantitative disease surveillance system for companion animals. Data from poison control centers, such as the Animal Poison Control Center (APCC), may be useful for conducting real-time surveillance in this population. In this study, we explored the suitability of APCC call data for early warning of toxicological incidents in companion animal populations by using a-priori knowledge of the melamine-related nephrotoxicosis outbreak. Patient and household-level information regarding possible toxicological exposures in dogs and cats reported to the APCC from 2005 to 2007, inclusive, were extracted from the APCC's AnTox database. These data were used to examine the impact of surveillance outcome, statistical methodology, analysis level, and call source on the ability to detect the outbreak prior to the voluntary recall issued by the pet food manufacturer. Retrospective Poisson temporal scan tests were applied for each combination of outcome, method, level, and call source. The results showed that month-adjusted scans using syndromic data may have been able to help detect the outbreak up to two months prior to the voluntary recall although the success of these methods varied across call sources. We also demonstrated covariate month-adjustment can lead to vastly different results based on the surveillance outcome and call source to which it is applied. This illustrates care should be taken prior to arbitrarily selecting a surveillance outcome and statistical model for surveillance efforts and warns against ignoring the impacts of call source or key covariates when applying quantitative surveillance methods to APCC call data since these factors can lead to very different results. This study provides further evidence that APCC call data may be useful for conducting surveillance in the US companion animal population and further exploratory analyses and validation studies are warranted.

A randomised controlled trial to reduce highest priority critically important antimicrobial prescription in companion animals

Robust evidence supporting strategies for companion animal antimicrobial stewardship is limited, despite frequent prescription of highest priority critically important antimicrobials (HPCIA). Here we describe a randomised controlled trial where electronic prescription data were utilised (August 2018-January 2019) to identify above average HPCIA-prescribing practices (n = 60), which were randomly assigned into a control group (CG) and two intervention groups. In March 2019, the light intervention group (LIG) and heavy intervention group (HIG) were notified of their above average status, and were provided with educational material (LIG, HIG), in-depth benchmarking (HIG), and follow-up meetings (HIG). Following notification, follow-up monitoring lasted for eight months (April-November 2019; post-intervention period) for all intervention groups, though HIG practices were able to access further support (i.e., follow-up meetings) for the first six of these months if requested. Post-intervention, in the HIG a 23.5% and 39.0% reduction in canine (0.5% of total consultations, 95% confidence interval, 0.4-0.6, P = 0.04) and feline (4.4%, 3.4-5.3, P < 0.001) HPCIA-prescribing consultations was observed, compared to the CG (dogs: 0.6%, 0.5-0.8; cats: 7.4%, 6.0-8.7). The LIG was associated with a 16.7% reduction in feline HPCIA prescription (6.1% of total consultations, 5.3-7.0, P = 0.03). Therefore, in this trial we have demonstrated effective strategies for reducing veterinary HPCIA prescription.

A systematic review of text mining approaches applied to various application areas in the biomedical domain

Purpose

This work shows the results of a systematic literature review on biomedical text mining. The purpose of this study is to identify the different text mining approaches used in different application areas of the biomedical domain, the common tools used and the challenges of biomedical text mining as compared to generic text mining algorithms. This study will be of value to biomedical researchers by allowing them to correlate text mining approaches to specific biomedical application areas. Implications for future research are also discussed.

Design/methodology/approach

The review was conducted following the principles of the Kitchenham method. A number of research questions were first formulated, followed by the definition of the search strategy. The papers were then selected based on a list of assessment criteria. Each of the papers were analyzed and information relevant to the research questions were extracted.

Findings

It was found that researchers have mostly harnessed data sources such as electronic health records, biomedical literature, social media and health-related forums. The most common text mining technique was natural language processing using tools such as MetaMap and Unstructured Information Management Architecture, alongside the use of medical terminologies such as Unified Medical Language System. The main application area was the detection of adverse drug events. Challenges identified included the need to deal with huge amounts of text, the heterogeneity of the different data sources, the duality of meaning of words in biomedical text and the amount of noise introduced mainly from social media and health-related forums.

Originality/value

To the best of the authors’ knowledge, other reviews in this area have focused on either specific techniques, specific application areas or specific data sources. The results of this review will help researchers to correlate most relevant and recent advances in text mining approaches to specific biomedical application areas by providing an up-to-date and holistic view of work done in this research area. The use of emerging text mining techniques has great potential to spur the development of innovative applications, thus considerably impacting on the advancement of biomedical research.

FasTag: Automatic text classification of unstructured medical narratives

Unstructured clinical narratives are continuously being recorded as part of delivery of care in electronic health records, and dedicated tagging staff spend considerable effort manually assigning clinical codes for billing purposes. Despite these efforts, however, label availability and accuracy are both suboptimal. In this retrospective study, we aimed to automate the assignment of top-level International Classification of Diseases version 9 (ICD-9) codes to clinical records from human and veterinary data stores using minimal manual labor and feature curation. Automating top-level annotations could in turn enable rapid cohort identification, especially in a veterinary setting. To this end, we trained long short-term memory (LSTM) recurrent neural networks (RNNs) on 52,722 human and 89,591 veterinary records. We investigated the accuracy of both separate-domain and combined-domain models and probed model portability. We established relevant baseline classification performances by training Decision Trees (DT) and Random Forests (RF). We also investigated whether transforming the data using MetaMap Lite, a clinical natural language processing tool, affected classification performance. We showed that the LSTM-RNNs accurately classify veterinary and human text narratives into top-level categories with an average weighted macro F1 score of 0.74 and 0.68 respectively. In the "neoplasia" category, the model trained on veterinary data had a high validation accuracy in veterinary data and moderate accuracy in human data, with F1 scores of 0.91 and 0.70 respectively. Our LSTM method scored slightly higher than that of the DT and RF models. The use of LSTM-RNN models represents a scalable structure that could prove useful in cohort identification for comparative oncology studies. Digitization of human and veterinary health information will continue to be a reality, particularly in the form of unstructured narratives. Our approach is a step forward for these two domains to learn from and inform one another.

Veterinary informatics: forging the future between veterinary medicine, human medicine, and One Health initiatives-a joint paper by the Association for Veterinary Informatics (AVI) and the CTSA One Health Alliance (COHA)

Objectives

This manuscript reviews the current state of veterinary medical electronic health records and the ability to aggregate and analyze large datasets from multiple organizations and clinics. We also review analytical techniques as well as research efforts into veterinary informatics with a focus on applications relevant to human and animal medicine. Our goal is to provide references and context for these resources so that researchers can identify resources of interest and translational opportunities to advance the field.

Methods and Results

This review covers various methods of veterinary informatics including natural language processing and machine learning techniques in brief and various ongoing and future projects. After detailing techniques and sources of data, we describe some of the challenges and opportunities within veterinary informatics as well as providing reviews of common One Health techniques and specific applications that affect both humans and animals.

Discussion

Current limitations in the field of veterinary informatics include limited sources of training data for developing machine learning and artificial intelligence algorithms, siloed data between academic institutions, corporate institutions, and many small private practices, and inconsistent data formats that make many integration problems difficult. Despite those limitations, there have been significant advancements in the field in the last few years and continued development of a few, key, large data resources that are available for interested clinicians and researchers. These real-world use cases and applications show current and significant future potential as veterinary informatics grows in importance. Veterinary informatics can forge new possibilities within veterinary medicine and between veterinary medicine, human medicine, and One Health initiatives.

An animal health example of managing and analyzing a large volume of data on a PC: Modeling body weight and age of over 13 million cats for explanatory and predictive purposes

Large amounts of animal health data are available to researchers, but are often stored in different formats and information silos. Analysis of this existing information can provide new insights into the health and welfare of animals and possibly reduce the need to collect additional data. The objective of this study was to develop a method of managing and analyzing large amounts of data on a personal computer that can be run within 24 h to limit the time and resources spent deploying models on larger servers. This paper describes an overall approach that makes use of existing methods for data acquisition and modeling, but adapts and combines them in a way that allows manipulation and analysis of large volumes of data on a PC. This included a total of five steps: removing errors; removing data points outside the scope of a specific hypothesis; creating descriptive statistics; developing explanatory and/or predictive models; and assessing the fit or accuracy of the models created. The approach was developed using electronic medical records for 19,416,753 feline patients from 3972 anonymized veterinary clinics in the United States and Canada, recorded between January 1981 and June 2016. Data regarding patient signalment (age, sex, breed, reproductive status) and body weight were extracted from the records and used to create linear regression models to describe body weight in cats of different ages, breeds, genders and reproductive status. Ordinary least squares linear regression and stochastic gradient descent linear regression were compared to determine their effectiveness and suitability for creating predictive models with large datasets, using 10 fold cross validation. This approach could be used to build workflows to create models to determine exploratory and predictive properties of health parameters for animals and people. The ability to work with large datasets on a PC or equivalent technology was demonstrated. Significant interactions were present among sex, reproductive status and age. A peak in weight occurred between 6 and 9 years depending on the sex, reproductive status and breed. The predictive ability of the two models was similar, with both producing a root mean square error of 1.45 and a mean absolute error of 1.09, and mean error that was approximately zero on the validation dataset.

Validation of text-mining and content analysis techniques using data collected from veterinary practice management software systems in the UK

Electronic patient records from practice management software systems have been used extensively in medicine for the investigation of clinical problems leading to the creation of decision support frameworks. To date, technologies that have been utilised for this purpose such as text mining and content analysis have not been employed significantly in veterinary medicine. The aim of this research was to pilot the use of content analysis and text-mining software for the synthesis and analysis of information extracted from veterinary electronic patient records. The purpose of the work was to be able to validate this approach for future employment across a number of practices for the purposes of practice based research. The approach utilised content analysis (Prosuite) and text mining (WordStat) software to aggregate the extracted text. Text mining tools such as Keyword in Context (KWIC) and Keyword Retrieval (KR) were employed to identify specific occurrences of data across the records. Two different datasets were interrogated, a bespoke test dataset that had been set up specifically for the purpose of the research, and a functioning veterinary clinic dataset that had been extracted from one veterinary practice. Across both datasets, the KWIC analysis was found to have a high level of accuracy with the search resulting in a sensitivity of between 85.3-100%, a specificity of between 99.1-99.7%, a positive predictive value between 93.5-95.8% and a negative predictive value between 97.7-100%. The KR search, based on machine learning, was utilised for the clinic-based dataset and was found to perform slightly better than the KWIC analysis. This study is the first to demonstrate the application of content analysis and text mining software for validation purposes across a number of different datasets for the purpose of search and recall of specific information across electronic patient records. This has not been demonstrated previously for small animal veterinary epidemiological research for the purposes of large scale analysis for practice-based research. Extension of this work to investigate more complex diseases across larger populations is required to fully explore the use of this approach in veterinary practice.

Review of Medical Decision Support and Machine-Learning Methods

Machine-learning methods can assist with the medical decision-making processes at the both the clinical and diagnostic levels. In this article, we first review historical milestones and specific applications of computer-based medical decision support tools in both veterinary and human medicine. Next, we take a mechanistic look at 3 archetypal learning algorithms—naive Bayes, decision trees, and neural network—commonly used to power these medical decision support tools. Last, we focus our discussion on the data sets used to train these algorithms and examine methods for validation, data representation, transformation, and feature selection. From this review, the reader should gain some appreciation for how these decision support tools have and can be used in medicine along with insight on their inner workings.

Proportion of nonsteroidal anti-inflammatory drug prescription in equine practice

BACKGROUND

There is little knowledge of the prescription of nonsteroidal anti-inflammatory drugs (NSAIDs) and whether their prescription varies between countries.

OBJECTIVE

To describe prescription practices of NSAIDs in equids in the United Kingdom (UK), United States of America (USA) and Canada.

STUDY DESIGN

Descriptive observational study.

METHODS

Free-text electronic medical records from 141,543 equids from 10 equine practices in the UK, 255,777 equids from 7 equine practices with 20 branches from the USA and 2 practices with 7 branches from Canada were evaluated. A validated text-mining technique was used to describe the proportion of equids prescribed NSAIDs at least once in these countries. The choice of NSAIDs in orthopaedic and colic cases was evaluated.

RESULTS

The prescription of NSAIDs is more common in the USA (42.4%) and Canada (34.2%) than in the UK (28.6%). Phenylbutazone and flunixin meglumine were the drugs mostly prescribed in all countries. While flunixin meglumine was most prescribed with colic cases in all countries, a proportion received phenylbutazone despite this drug being licensed for use only with musculoskeletal disease. Phenylbutazone was the most commonly prescribed drug in cases with orthopaedic disease followed by flunixin meglumine in all countries. Only a small proportion of cases received meloxicam, ketoprofen or firocoxib.

MAIN LIMITATIONS

The retrospective design might have resulted in an unknown number of incomplete records, particularly in the reporting of colic and orthopaedic disease. Although the data set is large, the relatively small number of practices recruited from each country may introduce bias.

CONCLUSIONS

Clinical practice can differ between countries although the influence of individual practitioners and practice-specific policy on apparent intercountry differences requires further research. Despite several other NSAIDs being available and a substantial effort being made to evaluate their efficacy, the prescription of NSAIDs other than phenylbutazone and flunixin meglumine remains rather limited.

Utility and potential of rapid epidemic intelligence from internet-based sources

OBJECTIVES

Rapid epidemic detection is an important objective of surveillance to enable timely intervention, but traditional validated surveillance data may not be available in the required timeframe for acute epidemic control. Increasing volumes of data on the Internet have prompted interest in methods that could use unstructured sources to enhance traditional disease surveillance and gain rapid epidemic intelligence. We aimed to summarise Internet-based methods that use freely-accessible, unstructured data for epidemic surveillance and explore their timeliness and accuracy outcomes.

METHODS

Steps outlined in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) checklist were used to guide a systematic review of research related to the use of informal or unstructured data by Internet-based intelligence methods for surveillance.

RESULTS

We identified 84 articles published between 2006-2016 relating to Internet-based public health surveillance methods. Studies used search queries, social media posts and approaches derived from existing Internet-based systems for early epidemic alerts and real-time monitoring. Most studies noted improved timeliness compared to official reporting, such as in the 2014 Ebola epidemic where epidemic alerts were generated first from ProMED-mail. Internet-based methods showed variable correlation strength with official datasets, with some methods showing reasonable accuracy.

CONCLUSION

The proliferation of publicly available information on the Internet provided a new avenue for epidemic intelligence. Methodologies have been developed to collect Internet data and some systems are already used to enhance the timeliness of traditional surveillance systems. To improve the utility of Internet-based systems, the key attributes of timeliness and data accuracy should be included in future evaluations of surveillance systems.

Translating Big Data into Smart Data for Veterinary Epidemiology

The increasing availability and complexity of data has led to new opportunities and challenges in veterinary epidemiology around how to translate abundant, diverse, and rapidly growing “big” data into meaningful insights for animal health. Big data analytics are used to understand health risks and minimize the impact of adverse animal health issues through identifying high-risk populations, combining data or processes acting at multiple scales through epidemiological modeling approaches, and harnessing high velocity data to monitor animal health trends and detect emerging health threats. The advent of big data requires the incorporation of new skills into veterinary epidemiology training, including, for example, machine learning and coding, to prepare a new generation of scientists and practitioners to engage with big data. Establishing pipelines to analyze big data in near real-time is the next step for progressing from simply having “big data” to create “smart data,” with the objective of improving understanding of health risks, effectiveness of management and policy decisions, and ultimately preventing or at least minimizing the impact of adverse animal health issues.

Validation of an Improved Computer-Assisted Technique for Mining Free-Text Electronic Medical Records

Background

The use of electronic medical records (EMRs) offers opportunity for clinical epidemiological research. With large EMR databases, automated analysis processes are necessary but require thorough validation before they can be routinely used.

Objective

The aim of this study was to validate a computer-assisted technique using commercially available content analysis software (SimStat-WordStat v.6 (SS/WS), Provalis Research) for mining free-text EMRs.

Methods

The dataset used for the validation process included life-long EMRs from 335 patients (17,563 rows of data), selected at random from a larger dataset (141,543 patients, ~2.6 million rows of data) and obtained from 10 equine veterinary practices in the United Kingdom. The ability of the computer-assisted technique to detect rows of data (cases) of colic, renal failure, right dorsal colitis, and non-steroidal anti-inflammatory drug (NSAID) use in the population was compared with manual classification. The first step of the computer-assisted analysis process was the definition of inclusion dictionaries to identify cases, including terms identifying a condition of interest. Words in inclusion dictionaries were selected from the list of all words in the dataset obtained in SS/WS. The second step consisted of defining an exclusion dictionary, including combinations of words to remove cases erroneously classified by the inclusion dictionary alone. The third step was the definition of a reinclusion dictionary to reinclude cases that had been erroneously classified by the exclusion dictionary. Finally, cases obtained by the exclusion dictionary were removed from cases obtained by the inclusion dictionary, and cases from the reinclusion dictionary were subsequently reincluded using Rv3.0.2 (R Foundation for Statistical Computing, Vienna, Austria). Manual analysis was performed as a separate process by a single experienced clinician reading through the dataset once and classifying each row of data based on the interpretation of the free-text notes. Validation was performed by comparison of the computer-assisted method with manual analysis, which was used as the gold standard. Sensitivity, specificity, negative predictive values (NPVs), positive predictive values (PPVs), and F values of the computer-assisted process were calculated by comparing them with the manual classification.

Results

Lowest sensitivity, specificity, PPVs, NPVs, and F values were 99.82% (1128/1130), 99.88% (16410/16429), 94.6% (223/239), 100.00% (16410/16412), and 99.0% (100×2×0.983×0.998/[0.983+0.998]), respectively. The computer-assisted process required few seconds to run, although an estimated 30 h were required for dictionary creation. Manual classification required approximately 80 man-hours.

Conclusions

The critical step in this work is the creation of accurate and inclusive dictionaries to ensure that no potential cases are missed. It is significantly easier to remove false positive terms from a SS/WS selected subset of a large database than search that original database for potential false negatives. The benefits of using this method are proportional to the size of the dataset to be analyzed.

Disease and pharmacologic risk factors for first and subsequent episodes of equine laminitis: A cohort study of free-text electronic medical records

Electronic medical records from first opinion equine veterinary practice may represent a unique resource for epidemiologic research. The appropriateness of this resource for risk factor analyses was explored as part of an investigation into clinical and pharmacologic risk factors for laminitis. Amalgamated medical records from seven UK practices were subjected to text mining to identify laminitis episodes, systemic or intra-synovial corticosteroid prescription, diseases known to affect laminitis risk and clinical signs or syndromes likely to lead to corticosteroid use. Cox proportional hazard models and Prentice, Williams, Peterson models for repeated events were used to estimate associations with time to first, or subsequent laminitis episodes, respectively. Over seventy percent of horses that were diagnosed with laminitis suffered at least one recurrence. Risk factors for first and subsequent laminitis episodes were found to vary. Corticosteroid use (prednisolone only) was only significantly associated with subsequent, and not initial laminitis episodes. Electronic medical record use for such analyses is plausible and offers important advantages over more traditional data sources. It does, however, pose challenges and limitations that must be taken into account, and requires a conceptual change to disease diagnosis which should be considered carefully.

Animal health syndromic surveillance: a systematic literature review of the progress in the last 5 years (2011-2016)

This review presents the current initiatives and potential for development in the field of animal health surveillance (AHSyS), 5 years on from its advent to the front of the veterinary public health scene. A systematic review approach was used to document the ongoing AHSyS initiatives (active systems and those in pilot phase) and recent methodological developments. Clinical data from practitioners and laboratory data remain the main data sources for AHSyS. However, although not currently integrated into prospectively running initiatives, production data, mortality data, abattoir data, and new media sources (such as Internet searches) have been the objective of an increasing number of publications seeking to develop and validate new AHSyS indicators. Some limitations inherent to AHSyS such as reporting sustainability and the lack of classification standards continue to hinder the development of automated syndromic analysis and interpretation. In an era of ubiquitous electronic collection of animal health data, surveillance experts are increasingly interested in running multivariate systems (which concurrently monitor several data streams) as they are inferentially more accurate than univariate systems. Thus, Bayesian methodologies, which are much more apt to discover the interplay among multiple syndromic data sources, are foreseen to play a big part in the future of AHSyS. It has become clear that early detection of outbreaks may not be the principal expected benefit of AHSyS. As more systems will enter an active prospective phase, following the intensive development stage of the last 5 years, the study envisions AHSyS, in particular for livestock, to significantly contribute to future international-, national-, and local-level animal health intelligence, going beyond the detection and monitoring of disease events by contributing solid situation awareness of animal welfare and health at various stages along the food-producing chain, and an understanding of the risk management involving actors in this value chain.

Prevalence, survival analysis and multimorbidity of chronic diseases in the general veterinarian-attended horse population of the UK

The average age of the global human population is increasing, leading to increased interest in the effects of chronic disease and multimorbidity on health resources and patient welfare. It has been posited that the average age of the general veterinarian-attended horse population of the UK is also increasing, and therefore it could be assumed that chronic diseases and multimorbidity would pose an increasing risk here also. However, evidence for this trend in ageing is very limited, and the current prevalence of many chronic diseases, and of multimorbidity, is unknown. Using text mining of first-opinion electronic medical records from seven veterinary practices around the UK, Kaplan-Meier and Cox proportional hazard modelling, we were able to estimate the apparent prevalence among veterinarian-attended horses of nine chronic diseases, and to assess their relative effects on median life expectancy following diagnosis. With these methods we found evidence of increasing population age. Multimorbidity affected 1.2% of the study population, and had a significant effect upon survival times, with co-occurrence of two diseases, and three or more diseases, leading to 6.6 and 21.3 times the hazard ratio compared to no chronic disease, respectively. Laminitis was involved in 74% of cases of multimorbidity. The population of horses attended by UK veterinarians appears to be aging, and chronic diseases and their co-occurrence are common features, and as such warrant further investigation.

Spatial-temporal clustering of companion animal enteric syndrome: detection and investigation through the use of electronic medical records from participating private practices

There is interest in the potential of companion animal surveillance to provide data to improve pet health and to provide early warning of environmental hazards to people. We implemented a companion animal surveillance system in Calgary, Alberta and the surrounding communities. Informatics technologies automatically extracted electronic medical records from participating veterinary practices and identified cases of enteric syndrome in the warehoused records. The data were analysed using time-series analyses and a retrospective space-time permutation scan statistic. We identified a seasonal pattern of reports of occurrences of enteric syndromes in companion animals and four statistically significant clusters of enteric syndrome cases. The cases within each cluster were examined and information about the animals involved (species, age, sex), their vaccination history, possible exposure or risk behaviour history, information about disease severity, and the aetiological diagnosis was collected. We then assessed whether the cases within the cluster were unusual and if they represented an animal or public health threat. There was often insufficient information recorded in the medical record to characterize the clusters by aetiology or exposures. Space-time analysis of companion animal enteric syndrome cases found evidence of clustering. Collection of more epidemiologically relevant data would enhance the utility of practice-based companion animal surveillance.

A practical approach to designing syndromic surveillance systems for livestock and poultry

The field of animal syndromic surveillance (SyS) is growing, with many systems being developed worldwide. Now is an appropriate time to share ideas and lessons learned from early SyS design and implementation. Based on our practical experience in animal health SyS, with additions from the public health and animal health SyS literature, we put forward for discussion a 6-step approach to designing SyS systems for livestock and poultry. The first step is to formalise policy and surveillance goals which are considerate of stakeholder expectations and reflect priority issues (1). Next, it is important to find consensus on national priority diseases and identify current surveillance gaps. The geographic, demographic, and temporal coverage of the system must be carefully assessed (2). A minimum dataset for SyS that includes the essential data to achieve all surveillance objectives while minimizing the amount of data collected should be defined. One can then compile an inventory of the data sources available and evaluate each using the criteria developed (3). A list of syndromes should then be produced for all data sources. Cases can be classified into syndrome classes and the data can be converted into time series (4). Based on the characteristics of the syndrome-time series, the length of historic data available and the type of outbreaks the system must detect, different aberration detection algorithms can be tested (5). Finally, it is essential to develop a minimally acceptable response protocol for each statistical signal produced (6). Important outcomes of this pre-operational phase should be building of a national network of experts and collective action and evaluation plans. While some of the more applied steps (4 and 5) are currently receiving consideration, more emphasis should be put on earlier conceptual steps by decision makers and surveillance developers (1-3).

Using informatics and the electronic medical record to describe antimicrobial use in the clinical management of diarrhea cases at 12 companion animal practices

Antimicrobial drugs may be used to treat diarrheal illness in companion animals. It is important to monitor antimicrobial use to better understand trends and patterns in antimicrobial resistance. There is no monitoring of antimicrobial use in companion animals in Canada. To explore how the use of electronic medical records could contribute to the ongoing, systematic collection of antimicrobial use data in companion animals, anonymized electronic medical records were extracted from 12 participating companion animal practices and warehoused at the University of Calgary. We used the pre-diagnostic, clinical features of diarrhea as the case definition in this study. Using text-mining technologies, cases of diarrhea were described by each of the following variables: diagnostic laboratory tests performed, the etiological diagnosis and antimicrobial therapies. The ability of the text miner to accurately describe the cases for each of the variables was evaluated. It could not reliably classify cases in terms of diagnostic tests or etiological diagnosis; a manual review of a random sample of 500 diarrhea cases determined that 88/500 (17.6%) of the target cases underwent diagnostic testing of which 36/88 (40.9%) had an etiological diagnosis. Text mining, compared to a human reviewer, could accurately identify cases that had been treated with antimicrobials with high sensitivity (92%, 95% confidence interval, 88.1%-95.4%) and specificity (85%, 95% confidence interval, 80.2%-89.1%). Overall, 7400/15,928 (46.5%) of pets presenting with diarrhea were treated with antimicrobials. Some temporal trends and patterns of the antimicrobial use are described. The results from this study suggest that informatics and the electronic medical records could be useful for monitoring trends in antimicrobial use.