PS 3-49 DEVELOPMENT OF A NATURAL HISTORY MODEL OF WEST NILE VIRUS INFECTIONS IN HUMANS

Tuesday, October 25, 2016
Bayshore Ballroom ABC, Lobby Level (Westin Bayshore Vancouver)
Poster Board # PS 3-49

Man Wah Yeung1, Emily Shing2, Mark Loeb3, David Naimark, MD, MSc, BSc4, Doug Sider2, Huaiping Zhu5 and Beate Sander, PhD2, (1)Health Quality Ontario, Toronto, ON, Canada, (2)Public Health Ontario, Toronto, ON, Canada, (3)McMaster University, Hamilton, ON, Canada, (4)University of Toronto, Toronto, ON, Canada, (5)York University, Toronto, ON, Canada
Purpose:

   To simulate the progression of West Nile virus (WNV) infections in humans over a lifetime time horizon in Ontario, Canada.

Method:

   We developed a microsimulation model to reflect the natural history of WNV human disease including neuroinvasive disease (ND) and non-neuroinvasive disease (non-ND). In symptomatic infection, individuals were classified as one of five clinical presentations, listed in order of increasing severity: non-ND, meningitis, encephalitis, meningoencephalitis and acute flaccid paralysis (AFP).

   Transition probabilities for hospitalizations, length of stay, other healthcare encounters, full recovery, persistence of long-term sequelae and WNV-related deaths were obtained from the literature. The acute phase (first year of infection) was modelled in weekly time intervals and the convalescent phase (period thereafter) was modelled in yearly intervals. We used population-based health administrative data to develop sociodemographic and disease profiles, reflecting the age, sex and comorbidity distributions of the Ontario population. Health utilities associated with clinical presentations were derived from patient-level data from a Canadian cohort of adults tested positive for WNV and followed for up to three years (Loeb 2008). Health utilities associated with long-term sequelae (e.g., difficulty with ambulation, myalgias, fatigue, depression) were obtained from the Tufts Cost-effectiveness Analysis Registry. Health outcomes were discounted at 5%. 

Result:

   The model is a good representation of WNV disease history: Mean age of infection was 55 years old. The distribution of clinical presentations was similar to patient-level data among individuals who developed ND: 7% AFP, 38% WNME, 33% WNE and 22% WNM. Rate of hospitalization among all individuals was 4.6%. The modelled number deaths were similar to observed data (case-fatalities: 8% ND, specifically 24% AFP, 11% WNME, 7% WNE, 1.6% WNM; and <1% non-ND). Individuals had a mean of 34 quality-adjusted life-years (QALYs); however, ND was associated with a mean 24 QALYs.

 Conclusion:

   The model synthesizes a broad range of data to aid in understanding the long-term disease burden and resource use associated with WNV. It lays the foundation to assess the comparative effectiveness and costs of WNV prevention strategies, including experimental human WNV vaccines currently undergoing early phase clinical trials.