Use of Public Health Intelligence for Disease Outbreaks (PHIDO) to Enhance Provincial Routine Reportable Disease Surveillance in Manitoba Weimin Hu, MBBS.

Use of Public Health Intelligence for Disease Outbreaks (PHIDO) to Enhance Provincial Routine

Reportable Disease Surveillance in Manitoba

Weimin Hu, MBBS MSc PhD

Public Health Officer, Canadian Public Health Service, Public Health Agency of Canada

CPHA Annual Conference, Toronto, May 26-29, 2014

Presentation Outline

• Introduction• Overview of routine reportable disease surveillance in Manitoba

» Biweekly epiVIEW surveillance reports• What is PHIDO - Public Health Intelligence for Disease

Outbreaks?» Added value PHIDO brings to provincial routine surveillance

• Provincial implementation of PHIDO» Conduct PHIDO assessment» Implement Knowledge Translation activities» Set provincial disease outbreak alert sensitivity» Develop provincial alert review process

• Enhanced routine reportable diseases surveillance report

Use of PHIDO to Enhance Routine Reportable Disease Surveillance in Manitoba

2

Use of PHIDO to Enhance Routine Reportable Disease Surveillance in Manitoba

3

Nu

mb

er

of

Re

po

rte

d C

as

es

Epidate

What is PHIDO?

• Public Health Intelligence for Disease Outbreaks, a standalone application, developed by BC Centre for Disease Control (BCCDC), can:

» Model patterns for the number of cases reported for a specific disease, within a specific region, age group, and or other factors, over time

» Evaluate the abnormality of a recently reported case count» Generate a disease outbreak alert to suggest further

investigations » Flag past point for group outbreak alerts as well

Use of PHIDO to Enhance Routine Reportable Disease Surveillance in Manitoba

4

How Does PHIDO Work?

• Historical data of case count by pre-defined time unit (daily, weekly, monthly, even quarterly) is used to model the expected cases at the same time unit

• Consideration for random process of disease occurrence, disease seasonality and long term trend

• All three components are combined together via a complicated statistical analysis to derive the expected case count at each pre-defined time unit

• Comparison between observed and expected case counts to flag potential outbreak alert

• Outbreak assessment and alert analysis outputs include graphs and text reports

Use of PHIDO to Enhance Routine Reportable Disease Surveillance in Manitoba

5

6

PHIDO’s Statistical Background

• Generalized Additive Modeling (GAM) for expected case counts modeling

t is time measured in days/weeks/months (1:k)

x1= sin(t*14π/365.25), x2= cos(t*14 π/365.25) - Seasonality

Estimate fi with LOWESS (locally weighted scatterplot smoothing)

• Assessments of the differences between the reported and the expected cases are based on a family of probability distributions for binomial data (Poisson, Binomial and Negative Binomial). The p-values at 0.05, 0.01, and 0.001 are set to flag low, medium and high alerts.

Use of PHIDO to Enhance Routine Reportable Disease Surveillance in Manitoba

),()()( 2121 xxftfLog tt

Use of PHIDO to Enhance Routine Reportable Disease Surveillance in Manitoba

7

Use of PHIDO to Enhance Routine Reportable Disease Surveillance in Manitoba

8

Use of PHIDO to Enhance Routine Reportable Disease Surveillance in Manitoba

9

Use of PHIDO to Enhance Routine Reportable Disease Surveillance in Manitoba

10

Implementation of PHIDO for Routine Surveillance

• Conducted Knowledge Translation of PHIDO among the key stakeholders involved provincial routine reportable disease surveillance

• Defined sensitivity levels of alert by type of diseases• Established a formal procedure for alert review and alert

reporting• Integrated into current biweekly epiVIEW surveillance report• Evaluated and incorporated feedback received:

» HIV» Influenza» Tuberculosis

Use of PHIDO to Enhance Routine Reportable Disease Surveillance in Manitoba

11

Use of PHIDO to Enhance Routine Reportable Disease Surveillance in Manitoba

12

QUESTIONS?

Use of PHIDO to Enhance Routine Reportable Disease Surveillance in Manitoba

13

Acknowledgements:

Ms. Michelyn Wood, MSc

MB/Sask Regional Coordinator, Canadian Public Health Program

Public Health Agency of Canada

Mr. Mark Saigeon

Manager, Canadian Public Health Service , Public Health Agency of Canada

Contact Information

Weimin Hu, MBBS MSc PhDPublic Health Officer, Canadian Public Health Program

Public Health Agency of Canada

[email protected] or [email protected]

Phone: 204-786-7292