Poison centers across the United States, Canada, Europe, Australia, and New Zealand have been long recognized for the direct patient care and educational programs that fulfill their core mandate [1-5]. The services provided by specialized pharmacists, nurses, and physicians at these centers typically include telephone consultations available 24 hours per day, every day of the year. Calls come from homes, workplaces, hospitals, and other health care facilities in urban, rural, and remote settings. Beyond their primary mandate to support the general public and health care professionals, poison centers also provide essential expertise in the field of toxicology through published articles and community outreach. Several studies in the United States have explored the direct impact of poison centers on avoided emergency room visits, physician visits, ambulance services, and other medical treatments [6-11]. A recent review found that every US $1 spent on poison centers translated to US $8 saved on unnecessary health care costs [12]. These studies also demonstrated that consultation with poison centers was associated with decreased hospital admissions for poisonings treated in emergency departments [7-11,13].

Public health concern over poison exposures is increasing because there are more potentially toxic products entering the consumer market [14-17], more pharmaceuticals being prescribed [18,19], new illicit drugs becoming available [19,20], and more toxins in the ambient environment [14,16,17]. Intentional and unintentional poison exposures are caused by hazardous mixtures of chemical, physical, and biological agents that affect humans and animals. Poisonings are the third leading cause of mortality from unintentional injury in Canada, behind motor vehicle crashes and falls [21]. One key tool in poison prevention is routine surveillance to identify long-term trends, short-term aberrations from those trends, and near-real-time changes in the occurrence of high risk cases. However, such surveillance is typically within the mandate of public health agencies and not within the mandate of poison centers. As such, strong partnerships are needed to fully harness the surveillance potential of data from poison centers.

Development of these partnerships has been prioritized in the United States, Europe, and New Zealand, where poison data have increased the capacity of public health surveillance to address morbidity and mortality from poisonings. Specifically, these systems have informed the response to food-borne illness [22,23], substance abuse [24], pharmaceutical misuse [25], and aquatic toxins [26]. They have also improved monitoring of covert threats [3] and mass public gatherings [27]. In the United States, much of this work is accomplished through the National Poison Data System (NPDS) [28], which provides nation-wide surveillance of exposures in near-real-time using a centralized database maintained by the American Association of Poison Control Centers. There is no equivalent infrastructure in Canada, and there is no routine review of the national data at predefined intervals [2]. Indeed, several recent reports comment on the reactive rather than proactive use of Canadian poison data to address specific issues, such as the meltdown of the nuclear reactors in Fukushima, Japan [29], deaths from contaminated ecstasy [30], and food-borne illnesses [31,32]. All of these examples highlight the potential for more systematic use of Canadian poison data for surveillance at the provincial and national scales.

There are 5 poison centers in Canada, all but one of which services more than one province or territory. In addition, 4 of these centers are associated with acute care facilities, but the British Columbia (BC) Drug and Poison Information Centre (DPIC) is located at the BC Centre for Disease Control (BCCDC) [2]. As a provincial public health agency, the BCCDC has the mandate and the resources necessary to conduct routine surveillance for a wide range of acute illnesses, chronic diseases, and environmental exposures. This mandate and these resources were extended to DPIC when it joined the BCCDC in 2011, making it possible to use the poison data for regular and systematic reporting to internal and external stakeholders. Since then, DPIC and the BCCDC have developed multiple surveillance systems related to different toxic exposures that operate on various time scales.

In late 2014, the research team launched the near-real-time alerting system for illness related to the consumption of shellfish (ASIRCS), which detects new cases within 1 hour of DPIC receiving the call. This was our first near-real-time system, and we highlight it here because (1) all cases of paralytic shellfish poisoning (PSP) are reportable under the Public Health Act of BC, (2) the automated system replaced a manual system with the same objectives, (3) shellfish-related illnesses are not limited to a particular demographic group or a particular time of year, and (4) it can easily be adapted to other applications or data from other poison centers.

Shellfish contaminated with bacteria, viruses, toxins, or some combination of the three can cause severe illness in those who consume them, and the incidence of such cases has been rising in BC over the past decade [33-35]. Examples of harmful contaminants include norovirus, Vibrio parahaemolyticus (VP), and different toxins that can accumulate in bivalve shellfish during harmful algal blooms.

Outbreaks related to all types have been reported in BC. In 2010, a norovirus outbreak caused 36 laboratory-confirmed illnesses among raw oyster consumers [33]. In 2015, an outbreak of VP caused acute gastroenteritis in 73 documented cases, which led to a ban on the sale of raw oysters in grocery stores and restaurants across the greater Vancouver area [34]. Estimates suggest there are approximately 350 unreported cases of infectious enteric illness for every reported case in BC [36]. In comparison with infectious agents, cases of toxic shellfish poisoning are rarer [37]. In 2011, an outbreak of diarrhetic shellfish poisoning (DSP) caused 60 illnesses in people who ate cooked mussels [35]. There have been 4 documented cases of PSP [38], which can cause paralysis, respiratory failure, and death [39]. Finally, concentrations of domoic acid, the toxin responsible for amnesic shellfish poisoning (ASP), have been increasing along the BC coast in recent years [40]. Although BC has not reported an outbreak related to ASP, the Canadian Food Inspection Agency notified the BCCDC of one case in 2016, and DPIC received a call from another suspected exposure involving three individuals in the same year.

The concept of reportable or notifiable diseases is fundamental to the practice of public health. Such diseases are considered to be of sufficient risk to the entire population that any cases diagnosed by doctors or laboratories much be reported to some central authority, often by law. These centralized data can then be used to identify and track outbreaks, and to inform public health decisions. Bacterial enteric illnesses are typically reported by a laboratory to a health authority following clinically relevant results, such as a positive test in stool or blood.

However, there are no clinical tests for shellfish biotoxins, and illnesses can only be confirmed if the toxin is detected in leftover products. This limitation means that most illnesses are reported as probable based on the symptoms collected by doctors. As there are no laboratory results, there is no systematic mechanism by which to report these illnesses to the responsible health authorities.

Sometimes there may be direct communication by doctors or individuals who are ill, but this is rare. Thus, it is challenging for health authorities to meet their duty to report cases of PSP or other shellfish toxins. Given that (1) DPIC receives calls about shellfish-related illnesses and is located within the BCCDC, and (2) the BCCDC has a responsibility to support mandatory reporting of PSP and other shellfish-related illnesses, the BCCDC decided to develop a system that would identify, isolate, and alert on all relevant cases. Here, we describe how ASIRCS works, the cases it identified in 2015 and 2016, and the public health actions related to those cases.

The province of BC is located on the west coast of Canada, and it has more than 25,000 km of coastline between its mainland and island areas [41]. These coastal waters have always been an important source of fish and shellfish for the First Nations of Canada [42]. In more recent history, they have played an important role in broader subsistence, recreational, and commercial fisheries. For example, BC oysters account for approximately 60% of all Canadian oyster production, which had a market value of CAN $27.3 million in 2013 [43]. The wholesale value of the entire BC shellfish industry was estimated at CAN $279 million in 2015 [44]. The 2015 population of BC was approximately 4.7 million residents, the majority of whom live in close proximity to the coast in greater Vancouver (2.3 million), on Vancouver Island (767,000), or elsewhere (186,000) [45].

The DPIC service has a dedicated 24×7 phone line that provides poison information to the general public and health care professionals, including doctors, nurses, and emergency medical personnel in the province of BC and the Yukon Territory. Poison information services are provided by pharmacists and nurses certified as poison information specialists. When someone calls the poison line, the responding DPIC specialist gathers the information necessary to assess the urgency of the concern, and then begins to obtain a complete history by following set guidelines. In some cases, the exposed individual makes the call, but in the other cases, the caller has not been exposed. Furthermore, a single caller may be calling about an incident in which multiple people were exposed.

In some situations, the single call will automatically generate a separate record for each exposed individual, meaning that the number of calls DPIC receives is smaller than the number of cases DPIC manages. Once there is a full understanding of the exposure, the DPIC poison specialists triage the exposed individual or individuals to a health care facility if needed or provide home or on-site management recommendations where appropriate.

Depending on the nature of the exposure, the poison specialists routinely use their clinical experience and judgment along with other sources of toxicological information such as the Poison Management Manual [46] and Micromedex Solutions [47] to manage cases. Severe or unusual exposures may require follow-up calls to the treating health care facility until symptoms have subsided or life-threatening conditions have been resolved. All follow-up information is included in the single record for each case.

Since October 2011, DPIC has maintained records of all cases using the WBM software Visual Dotlab Enterprise (VDLE) [48], an electronic system specifically designed for use by poison centers. Like many other poison centers, DPIC staff use SAP Crystal Report XI Developer [49] to generate in-house summaries of data in VDLE, but this system lacks the analytic capacity to support routine surveillance. All of the DPIC surveillance systems, including ASIRCS, have been built in the R statistical computing environment [50], using the RODBC package [51] to interface with the VDLE database. Working with the data in R allows easy and flexible processing, manipulation, analysis, tabulation, plotting, and reporting on data collected by DPIC.

Every hour on the hour, ASIRCS connects to the VDLE database and extracts all cases from the past 24 hours. The system uses a 24-hour period rather than a 1-hour period because it sometimes takes several hours for a case to be closed within VDLE. Case information is then extracted and merged from the relevant raw data tables before data cleaning.

The multistep procedure of data cleaning is challenging because it requires the transcription of information captured by the DPIC poison specialist into a standardized tabular format. For example, it is particularly challenging to extract relevant information from free-text fields in VDLE, where poison specialists document history, assessment, and recommendations as in a medical chart. Once all the data have been extracted and merged, ASIRCS scans the data for cases related to the consumption of shellfish based on criteria established by BCCDC food safety specialists.

An alert is generated when the following criteria are met:

When new cases are detected, ASIRCS generates an automated report tailored to the needs of the BCCDC food safety specialists (Multimedia Appendix 1). The file is saved in a secure location to protect personal privacy, and ASIRCS sends an automatic email to system users to let them know that an alert has been generated. The BCCDC food safety specialists review the alerted case or cases and use the information along with their expertise to assess whether the case or cases should be referred to the most responsible of the 5 regional health authorities for potential reporting. Once referred, the environmental health officers contact the patient or patients to collect further information about the case or cases, provided that consent and contact information for such follow-up was given at the time of the call to DPIC.

Beta testing of ASIRCS started in late 2014, and we now have complete data for both 2015 and 2016. Here, we have reviewed all cases identified by the system over these 2 years and summarized them with respect to the nature of the exposure and public health follow-up. Specifically, we have identified those that were referred to the regional health authorities, and we have identified potential exposures to each group of shellfish toxins: PSP, ASP, or DSP. The cases were categorized into these 3 groups based on the reported symptoms matching each of the shellfish toxin exposure: numbness and tingling are indicative of PSP; headache, confusion, and disorientation are indicative of ASP; and diarrhea and abdominal cramps are indicative of DSP when symptom onset is <10 hours. Norovirus, VP, and other enteric illnesses may be considered when symptom onset is >10 hours.

Finally, we retrospectively checked all VDLE records from the same 2 years to evaluate whether we missed any cases that should have generated alerts. First, we applied the same criteria described above to all cases in the VDLE database and compared the results with the list of alerted cases. Second, we added the keyword “seafood” to the criteria above to evaluate whether extension of the keyword list would provide improved results.