Introducing the Flu and Other Respiratory Viruses Research (FOREVER) Cohort, a laboratory partnership for data sharing

Transcription

1 Introducing the Flu and Other Respiratory Viruses Research (FOREVER) Cohort, a laboratory partnership for data sharing JEFF KWONG PHO GRAND ROUNDS JULY 25,

2 Financial disclosures No financial conflicts of interest to disclose 2

3 Study team Jonathan Gubbay Aaron Campigotto Tim Karnauchow Kevin Katz Allison McGeer Dayre McNally David Richardson Susan Richardson Andrew Simor Marek Smieja George Zahariadis Hannah Chung Michael Campitelli Sarah Buchan Kevin Schwartz Natasha Crowcroft Laura Rosella Michael Jackson Dennis Ko Funding: Canadian Institutes of Health Research (MOP ) 3

4 Learning objectives After the session, participants will be able to: Describe the creation of the FOREVER Cohort Appreciate the use of routinely collected laboratory and health administrative data for evaluating influenza vaccine effectiveness Discuss some other applications of these data 4

5 Annual burden of illness from influenza in Canada 3,500 deaths 12,000 hospitalizations 160,000 emergency department visits 900,000 physician office visits 2,300,000 work absences 3,500,000 symptomatic infections 5

6 Health burden of respiratory viruses Kwong JC, et al. Ontario Burden of Infectious Diseases Study (ONBOIDS): An OAHPP/ICES Report

7 Protecting yourself and others from influenza and other respiratory viruses Hand hygiene Respiratory etiquette Social distancing Masks/respirators Influenza vaccination 7

8 Influenza vaccine coverage in Canada, % 34% 35% Overall: 31% (>11 million Canadians) 32% 31% 30% 30% 33% 24% 27% 35% 37% 45% Buchan SA and Kwong JC. Trends in influenza vaccine coverage and vaccine hesitancy in Canada, 2006/07 to 2013/14: results from cross-sectional survey data. CMAJ Open 2016;4(3):E

9 Creating the Flu and Other Respiratory Viruses Research (FOREVER) Cohort Linked respiratory virus testing data for the period May 2009 to May 2014 from 11 PHO and 8 academic hospital labs: Children s Hospital of Eastern Ontario Hospital for Sick Children Mount Sinai Hospital North York General Hospital St Joseph s Healthcare Hamilton Sunnybrook Health Sciences Centre University Health Network William Osler Health System (London Health Sciences Centre) Overall linkage proportion: 97.5% 9

10 Laboratories contributing to the FOREVER Cohort Red = PHO lab Blue = hospital lab 10

11 Respiratory viruses tested for Influenza A Influenza B Respiratory syncytial virus (RSV) Adenovirus Coronavirus Enterovirus/rhinovirus Parainfluenza virus Human metapneumovirus Bocavirus 11

12 Reason for testing and laboratory testing methods Reason for testing: Routine clinical care Outbreak investigations Research Laboratory testing methods: Polymerase chain reaction (PCR) (monoplex & multiplex) Direct fluorescent antibody staining (DFA) Viral culture Enzyme immunoassay (EIA) rapid antigen tests 12

13 Participating labs, viruses tested, methods used 13

14 Care Providers ICES Physician Data Base physnum presc_i Health Services Administrative Data OHIP Physician Claims Ontario Drug Benefit Claims ikn ikn Population & ICES Data Holdings Special Populations (restricted access) Health Care Facilities Master Numbering System/Hosp_DB aminst inst inst facnum inst inst National Ambulatory Care Reporting System Hospital Discharge Abstracts Ontario Home Care System Chronic Care Reporting System National Rehabilitation Reporting System Levels of Care Reporting System ikn ikn ikn ikn ikn ikn Geography REGISTERED PERSONS DATABASE Postal Code Conversion/ Geographical Population Estimates Canada Census Profiles ikn ikn ikn ikn ikn Cardiac Care Network Registry of the Canadian Stroke Network Ontario Cancer Registry Chronic Condition Cohorts 3rd Party/ Abstracted/Primary Collected Data Survey Data Canadian Community Health Survey Research Data Sets Sample of ICES Data Holdings 14

15 Using ICES data to assign healthcare setting 9% 6% 6% 14% ICU Ward ED 20% 45% MD office LTCF Missing 15

16 Using the FOREVER Cohort to assess influenza vaccine effectiveness 16

17 Background Annual influenza vaccination recommended for older adults Systematic review of 35 test-negative design studies in community-dwelling adults aged 60 years: Vaccine match: VE=52% (95%CI, 41%-61%) Vaccine mismatch: VE=36% (95%CI, 22%-48%) But evidence that influenza vaccines reduce laboratoryconfirmed serious outcomes in older adults is relatively sparse Darvishian M, et al. Effectiveness of seasonal influenza vaccine in community-dwelling elderly people: a meta-analysis of testnegative design case-control studies. Lancet ID 2014;14:

18 Test-negative design Has become the preferred method for assessing influenza VE (when RCTs not possible) Estimates VE against medically-attended laboratoryconfirmed influenza Like a nested case-control study: among those who seek medical care for ARI/ILI, those who test positive for influenza are cases and those who test negative are controls Removes confounding from health-seeking behaviour 18

19 Typical test-negative design studies Case definition for ILI/ARI Symptom onset 7 days Tested by PCR Sullivan SG, et al. Potential of the test-negative design for measuring influenza vaccine effectiveness: a systematic review. Expert Review of Vaccines 2015;13(12):

20 Recommendations for VE studies using the TND VE analyses based on a convenience sample of clinical diagnostic tests could be biased and should be avoided Belongia EA, et al. Variable influenza vaccine effectiveness by subtype: a systematic review and meta-analysis of testnegative design studies. Lancet ID 2016;16(8):

21 Confirming the appropriateness and validity of using the FOREVER Cohort to estimate influenza VE 1. Test assumptions of the TND 2. Estimate VE under various conditions 3. Compare FOREVER VE estimates with the literature 21

22 Assumption 1: Influenza vaccination and non-influenza viruses Influenza vaccination should not be associated with infections by non-influenza viruses Vaccine coverage should be similar between the control group (i.e., influenza-negative individuals) and the source population (i.e., all tested individuals) De Serres G, et al. The test negative design: validity, accuracy and precision of vaccine efficacy estimates compared to the gold standard of randomized placebo-controlled trials. Euro Surveillance 2013;18(37): Jackson ML & Nelson JC. The test-negative design for estimating influenza vaccine effectiveness. Vaccine. 2013;31(17):

23 Assumption 1: Influenza vaccination and non-influenza viruses Source population Tested for influenza Should have similar vaccine coverage Cases Influenzapositive Influenzanegative Controls Other viruspositive Pannegative 23

24 Assumption 1: Influenza vaccination and non-influenza viruses Restricted to community-dwelling adults aged >65 years tested by multiplex PCR (panel of influenza + 5 other viruses) Healthcare settings: Hospitals Emergency departments Physician offices Diagnostic codes: ARI-coded (substitute for case definition) Non-ARI-coded 24

25 Assumption 2: Influenza vaccination and testing for influenza Testing for influenza should not be influenced by vaccination status Orenstein EW, et al. Methodologic issues regarding the use of three observational study designs to assess influenza vaccine effectiveness. Int J Epidemiol. 2007;36(3): Ferdinands JM, et al. Influenza vaccination status is not associated with influenza testing among children: Implications for observational studies of vaccine effectiveness. Vaccine. 2011;29(10):

26 Assumption 2: Influenza vaccination and testing for influenza Restricted to community-dwelling adults aged >65 years Extracted healthcare encounters during influenza season Healthcare settings: Hospitals Emergency departments Physician offices Diagnostic codes: ARI-coded (substitute for case definition) Non-ARI-coded 26

27 Estimating VE under various conditions Restricted to community-dwelling adults aged >65 years VE = (1 aor) x100%, adjusted for age, sex, calendar month of test, presence of any comorbidity Healthcare setting: Outpatients and inpatients combined Outpatients Inpatients Laboratory testing methods: Tested by any laboratory method Tested by PCR 27

28 VE in inpatients vs. outpatients Compared 25 pairs of estimates Pooled difference = 2% (95% CI 12%, 10%) Feng S, et al. Influenza vaccine effectiveness by test-negative design Comparison of inpatient and outpatient settings. Vaccine 2016;34:

29 Including specimens tested by non-pcr methods Impact of test sensitivity on VE is far less than test specificity Other available testing methods are not as sensitive as PCR, but have comparable specificity Effect of decreasing sensitivity on VE Effect of decreasing specificity on VE Jackson ML & Rothman KJ. Effects of imperfect test sensitivity and specificity on observational studies of influenza vaccine effectiveness. Vaccine 2015;33:

30 Estimating VE under various conditions Applying restrictions to simulate typical TND studies A. Inpatients (no restrictions) B. ARI-coded hospitalizations C. Non-outbreak-related specimens D. Documented symptom onset 7 days E. ARI-coded + non-outbreak F. ARI-coded + non-outbreak + symptom onset 7 days G. Correcting for exposure misclassification Schwartz KL, et al. Using physician billing claims from the Ontario Health Insurance Plan to determine individual influenza vaccination status: an updated validation study. CMAJ Open 2016;4(3):E Fox MP, et al. A method to automate probabilistic sensitivity analyses of misclassified binary variables. Int J Epi 2005;34:

31 Additional influenza VE studies (in progress/future) Influenza VE in: Young children (6-59 months) Pregnant women Older adults with: COPD Asthma Cancer Diabetes Cardiovascular disease Chronic kidney disease VE of LAIV vs. IIV in children aged 2-17 years (Alberta) Impact of repeated vaccination on VE in young children (6-59 months) Intraseason waning of VE Impact of statins on VE 31

32 Risk of acute myocardial infarction after laboratory-confirmed influenza infection 32

33 Background Coronary artery disease remains number one cause of mortality globally Long-recognized temporal associations between influenza epidemics and CV mortality, supported by epidemiological studies of association between ARI/ILI and acute CV events Pathophysiological mechanisms: systemic inflammation, vasoconstriction, platelet activation, endothelial dysfunction, increased metabolic demand Only 1 study (case-control) has examined association between laboratory-confirmed influenza infection (using serology) and AMI 33

34 Self-controlled case-series design Exposure Outcome 34

35 Protecting yourself and others from influenza and other respiratory viruses to reduce the risk of AMI Hand hygiene Respiratory etiquette Social distancing Masks/respirators Influenza vaccination 35

36 Take-home messages Linking laboratory data to health administrative data represents a relatively inexpensive and efficient way to create a large cohort of individuals tested for an infectious agent Use of routinely collected data for TND studies is likely appropriate and produces estimates similar to typical TND studies Efforts to reduce the burden of AMI should include measures to prevent respiratory infections Data linkage opens up a plethora of opportunities for infectious diseases research 36

37 Questions? 37