This study contributes to a small but growing body of literature aimed at documenting whether and how syndromic data is linked with public health action, and additionally describes capacities to use these data. In the context of the recent pandemic in Ontario, this study shows that data from syndromic surveillance systems had limited application in decision-making regarding selected public health actions. Many of these decisions were instead driven by other logistical and contextual factors (e.g. vaccine availability and MOHTLC recommendations). Other influences on decision-making for both SSUs and NSUs were traditional surveillance from laboratory data and communications with stakeholders.
Syndromic surveillance data were most frequently used for communications and messaging, both internally within an organization and externally with stakeholders, partners and the media. Specifically, syndromic data were used in surveillance bulletins to communicate A(H1N1)pdm09 activity, improve risk communication and support recommendations, and in the operation (e.g. opening, closing and placement) of influenza assessment centres. Additionally, syndromic data were valued for monitoring virus activity in the community and providing credibility and confidence to support decisions and recommendations. These uses reflect a reactive approach to the impact of A(H1N1)pdm09 activity on the health care system, rather than a proactive approach to early identification and action (e.g. advanced set up of control systems such as assessment centres before health care systems were pressured).
Few studies have examined the public health actions taken in response to syndromic data. In the US, case studies have found that even when received in a timely manner, syndromic surveillance did not have an influence on public health actions for seasonal influenza [9
]. In contrast, data from England and Wales’ national telephone health advice helpline, NHS Direct, have been used to track seasonal influenza and to communicate risk of adverse events and provide reassurance to the public following a fuel explosion [17
US studies have also reported the use of syndromic data for monitoring of disease activity and communicating surveillance findings to stakeholders and the community [9
]. Uncertainty about the ability of syndromic surveillance to detect outbreaks has led to its greater utility for situational awareness regardless of the method of detection, particularly for monitoring influenza activity and its public health impact [19
]. The value of syndromic surveillance for monitoring disease activity in the US is reflected by the finding that 98% (40/41) of respondents (from state, territorial and large, local jurisdictions) indicated that they planned to use syndromic surveillance to monitor the impact of pandemic influenza [19
In this study population, syndromic surveillance was most used for monitoring A(H1N1)pdm09 activity and its impact on the health care system, as well as supporting communications and messaging, rather than use for its intended purpose of early outbreak detection. SSUs valued the information about community disease activity that syndromic data provided and the reassurance and confidence it provided to decision-makers. However, if not linked to action or contributing to the prevention and control of adverse health related events, the utility and benefits of syndromic surveillance are unclear [5
Given the use of syndromic surveillance to support and reinforce public health messages, this study raises questions about the value of syndromic surveillance as an adjunct to traditional influenza surveillance systems. It is unclear how many surveillance systems are needed if these systems are most beneficial for situational awareness and providing supportive evidence to increase credibility and confidence in decisions previously made. Further familiarity with syndromic surveillance systems and development of algorithms for generating and responding to alerts may improve its utility for decision-making. Even with such enhancements, the utility of syndromic surveillance data may continue to be constrained if other systemic barriers are not addressed. Determining how and under which conditions the utility of syndromic surveillance can be maximized requires further study.
The strengths of this study include good representation (82%) from decision-makers and data users across Ontario’s public health departments. The use of semi-structured interviews gave participants flexibility in their responses and allowed us to obtain a richer understanding of system experiences. Having decision end points and outcomes provided objectivity to evaluating the usefulness of syndromic data in supporting public health actions during the pandemic.
One important limitation of this study’s findings is the high level of variation we noted in the operation of syndromic surveillance systems. With 27% of systems in operation for less than two years, many users indicated a lack of experience or ability to use historical trends to define triggers for action, which likely contributed to its limited use for decision-making. These findings are similar to a study of eight US states where only 48% of SSUs had written response protocols, and where the lack of a systematic process for designing protocols and few available information resources for response were identified [14
]. For systems in place for longer than two years, however, the reasons for not having triggers are unknown. Additionally, as this study asked about use of syndromic data for specific decision end points, whether syndromic surveillance had any additional indirect impacts on decision-making more generally (beyond providing reassurance and confidence) remains unanswered.
As respondents were limited to those from local, provincial and federal public health agencies, we also have not captured how syndromic data were used by data providers (e.g. hospitals, schools). Another study limitation includes the potential for recall bias as interviews were conducted almost a year after the pandemic peak. Although participants were assured of confidentiality, other possible biases include providing responses that are likely to be viewed more favourably, attract further funding, or as socially desirable, especially when much time and effort was invested in collecting and analyzing syndromic data.