Developing a harmonized heat warning and information system for Ontario: a case study in collaboration.

BACKGROUND: Heat wave early warning systems help alert decision-makers and the public to prepare for hot weather and implement preventive actions to protect health. Prior to harmonization, public health units across Ontario either used independent systems with varying methodologies for triggering and issuing public heat warnings or did not use any system. The federal government also issued heat warnings based on different criteria. During heat events, adjacent public health units in Ontario and the federal government would routinely call heat warnings at different times with separate public messages, leading to confusion. This article describes the collaborative process and key steps in developing a harmonized Heat Warning and Information System (HWIS) for Ontario. SETTING: Public health units across Ontario, Canada, collaborated with the federal and provincial government to develop the harmonized HWIS for Ontario. INTERVENTION: In 2011, stakeholders identified the need to develop a harmonized system across Ontario to improve heat warning services, warning criteria, and health messaging. Through a 5-year process facilitated by a non-governmental organization, the three levels of government collaborated to establish the Ontario HWIS. OUTCOMES: The province-wide HWIS was implemented in 2016 with the Ontario Ministry of Health and Long-Term Care's release of the harmonized HWIS Standard Operating Practice, which outlined the notification and warning process. IMPLICATIONS: The lessons learned could help spur action in other provinces and jurisdictions internationally in the development of similar health evidence-based warning systems, including in particular those for protecting public health during extreme heat events.

Mixture Markov regression model with application to mosquito surveillance data analysis.

A mixture Markov regression model is proposed to analyze heterogeneous time series data. Mixture quasi-likelihood is formulated to model time series with mixture components and exogenous variables. The parameters are estimated by quasi-likelihood estimating equations. A modified EM algorithm is developed for the mixture time series model. The model and proposed algorithm are tested on simulated data and applied to mosquito surveillance data in Peel Region, Canada.