Ontario

1
Ontarios Experience with a Universal Influenza
Immunization Program (UIIP)

• Doug Manuel, MD MSc FRCPC
• Scientist
• Jeff Kwong, MD MSc CCFP
• Research Fellow
• October 25, 2005

2
Outline

• Background
• Effect of UIIP on vaccination rates
• Effect of UIIP on hospitalizations
• Discussion conclusions

3
A brief geography lesson
4
Population by province/territory as of July 1,
2004
30K
40K
30K
0.5M
7M
3M
12M
1M
0.1M
4M
1M
1M
0.7M
5
Influenza vaccination in Canada

• All health care services are publicly funded and
  delivered, but programs vary by province
• Ontario, pre-2000
• 1988 Targeted program initiated to provide
  influenza vaccination free for those at high risk
  of complications from influenza (elderly 65,
  those with chronic medical conditions)
• 1993 Program expanded to cover patient-care
  staff of long term care facilities (LTCF)
• 1999 Program expanded to cover ALL health care
  workers

6
Ontario launches UIIP in 2000

• First large-scale program to provide free
  influenza vaccination to entire population aged 6
  months or older
• Stated goals to decrease influenza-related
  morbidity and mortality, and to decrease ER
  overcrowding
• All other provinces chose to maintain their
  targeted vaccination programs

7
Some details about the UIIP

• Vaccines (trivalent inactivated vaccine)
  purchased centrally by Ministry of Health and
  Long-Term Care from 2 manufacturers
  Sanofi-Pasteur and Shire Biologics
• Vaccines delivered at MD offices, LTC facilities,
  hospitals, public health units, workplace
  clinics, pharmacies, schools, community centers,
  shopping malls, etc.

8
Locations where people were vaccinated in
Ontario, 2000-01
9
More details about the UIIP

• Extensive communications campaign by provincial
  government and local public health units to
  promote UIIP, including TV/radio/print
  advertising, newsletters, mailings, billboards,
  web sites, etc.
• Response has been generally favorable by the
  public and those involved in delivery of vaccines

10
Estimated costs of the UIIP

• Total program cost 42M CDN (2003-04)
• Vaccine purchase (54)
• Vaccine delivery administration (34)
• Communications (12)
• 1CDN 0.85US as of Oct 17, 2005

11
Effect of Ontarios UIIP on vaccination rates
12
Research Questions

• Did introduction of Ontarios UIIP in 2000 lead
  to an increase in vaccination rates, compared to
  the 9 provinces that maintained targeted
  immunization programs?
• If so, which population subgroups benefited the
  most?

13
Data sources

• National Population Health Survey (NPHS)
• Canadian Community Health Survey (CCHS)
• Both conducted by Statistics Canada
• Cover the household population
• Exclude members of the Canadian Forces, native
  reserves, and some remote areas, those living in
  institutions (e.g., nursing homes, prisons)

14
Population health surveys
NPHS 1996/97 CCHS 1.1 2000/01 CCHS 2.1 2003
Data collection Jun 1996 to Aug 1997 Jun to Aug 2001 (Q4) Jan to Dec 2003
Response rate 83 85 81
Sample 73,402 35,187 133,026
Weighted population 24.6 million 25.9 million 26.5 million
15
Unpublished results to be presented
16
Summary of findings

• Influenza vaccination rates increased across
  Canada between 1996 2003
• Introduction of UIIP in Ontario associated with a
  significant increase in coverage rate of overall
  population aged 12 years compared to other
  provinces, with the increase accounted for by
  those aged 12-64 years

17
Median coverage rates in LTC facility residents
staff, hospital staff

• Since 1999, coverage rates as of Nov 15 provided
  annually by facilities to public health officials

1999-00 2000-01 2001-02 2002-03 2003-04
Residents 93 95 96 95 95
LTC Staff 86 90 86 82 84
Hosp Staff N/A 63 51 44 55
Includes those vaccinated in Dec and Jan
18
Effect of Ontarios UIIP on hospitalizations
19
Research Questions

• Did introduction of Ontarios UIIP in 2000 lead
  to a reduction in influenza-related
  hospitalizations, compared to the 9 provinces
  that maintained targeted immunization programs?
• If so, which population subgroups benefited the
  most?

20
Methods

• Study design Interrupted time series with
  concurrent controls
• Study population and setting All residents of
  10 Canadian provinces, September 1993 to March
  2004

21
Hospitalization data

• Obtained from Hospital Morbidity Database
• Included admissions with the following conditions
  listed as 1 of the first 5 diagnoses

Diagnoses (Dx) ICD-9 Codes ICD-10 Codes
Pneumonia Influenza (PI) 480-487 J10-18
Acute Respiratory Diseases (ARD) 460-466 J00-06, J20-22
Chronic Obstructive Pulmonary Disease (COPD) 490-492, 496 J40-44
22
Definition of influenza season

• From October to May, starting when each week
  accounted for 5 of the seasons total number
  of influenza virus isolates for 2 weeks and
  ending when the influenza isolates accounted for
  lt 5 for 2 weeks

Adapted from Izurieta HS, et al. NEJM 2000
342(4)232-9
23
Viral surveillance data

• Respiratory virus detections
• Weekly percentage of tests positive for influenza
  A, influenza B, RSV
• Predominant influenza subtypes
• Subtype considered predominant if detections
  accounted for 20 of the seasons isolates
• Vaccine antigenic match
• Compared mismatch between circulating strains and
  vaccine strains
• Good match lt 20 circulating strains mismatched
• Fair match 20-50 circulating strains mismatched
• Poor match gt 50 circulating strains mismatched

24
Statistical analysis

• Poisson regression models
• Run separately for each condition and province,
  for 7 age groups 0-4, 5-19, 20-49, 50-64, 65-74,
  75-84, 85
• Accounted for numerous covariates
• Used generalized estimating equations to control
  for autocorrelation AR(1)

Adapted from Thompson, et al. JAMA 2004
292(11)1333-40
25
Poisson regression model

• Y a exp (ß0 ß1UIIP_flu ß2sexagegrp
• ß3FluA ß4FluB ß5RSV
• ß6A(H1N1) ß7A(H3N2) ß8B
• ß9match ß10ICD-10
• ß11t ß12t2 ß13t3
• ß14-16sin(2tkp/52) ß17-19cos(2tkp/52)
• where k1, 2, 3 e)

26
Poisson model terms

• Y weekly number of condition-specific
  hospitalizations for a given province-, sex-, and
  age group-stratum
• a log of province-, sex- and age group-specific
  annual population size
• UIIP_flu RR of hospitalizations during
  influenza seasons after 2000 vs. before
• sexagegrp sex age group interaction
• FluA/FluB/RSV weekly of regional tests
  positive for influenza A/B, RSV
• A(H1N1)/A(H3N2)/B predominant influenza
  subtype(s) for a season 1/0
• match vaccine antigenic match G/F/P
• ICD-10 introduction of ICD-10
• t, t2, t3 time trend terms tweek number
  divided by 52
• sin/cos seasonal trend terms

27
Statistical analysis contd

• Compared RR estimates of change in influenza
  season-associated hospitalizations over time for
  Ontario with pooled estimates for other provinces
  combined
• Pooled RR estimates for separate age groups to
  estimate effect on overall pop.
• Pooled RR estimates for separate conditions to
  estimate combined effect on the 3
  influenza-related conditions

28
Unpublished results to be presented
29
Discussion Conclusions
30
Limitations

• Data quality concerns with health administrative
  data (coding validity/reliability)
• No vaccine coverage data on children lt 12 years
  or institutionalized elderly (outside ON)
• Provincial-level analysis may have blurred
  regional variations in timing and severity of
  influenza epidemics and hospitalization rates
• Ecological study design susceptible to
  ecological fallacy
• Unmeasured confounders

31
Lessons learned

• Implementation of a universal influenza
  vaccination program is feasible
• Clear increases in vaccination rates observed in
  younger age groups, and increases generally
  sustained
• Modest reductions in influenza-related
  hospitalizations observed in groups with larger
  increases in coverage rates
• Suggests direct benefit of influenza vaccination
• Uncertain about indirect benefits (herd immunity)

32
Critical information gaps

• Suboptimal data on individual-level vaccination
  status (no immunization registry)
• Effect of UIIP on other outcomes (outpatient MD
  visits, ER services, mortality) to be examined by
  early 2006
• Effect of UIIP on school and workplace
  absenteeism should also be assessed
• No data on vaccination rates in those lt 12 yrs
• Economic evaluation based on empirical outcome
  data needed

33
Future studies

• Examine other outcomes
• Repeat time series analyses with additional data
  (i.e., more influenza seasons)
• Cross-sectional study to examine the relationship
  between regional variations in vaccination rates
  and outcome rates in post-UIIP Ontario (to look
  for a dose-response relationship)

34
Acknowledgements

• ICES
• Therese Stukel, Jenny Lim, Laura Fazio
• Statistics Canada
• Helen Johansen, Christie Sambell
• Public Health Agency of Canada
• Peter Zabchuk
• CDC
• Bill Thompson, David Shay
• Others
• Ross Upshur, Allison McGeer, Susan Tamblyn, Irfan
  Dhalla

35
Financial support

• This research was supported by a grant from the
  Public Health Agency of Canada
• Doug Manuel is supported by a Career Scientist
  award from Ontarios Ministry of Health and
  Long-Term Care
• Jeff Kwong is supported by a CIHR Fellowship
  award from the Canadian Institutes of Health
  Research

36
Questions?