Improving Flu Vaccination Rates for Children with Chronic Conditions CHEAR Unit, Division of General Pediatrics, University of Michigan Improving Flu Vaccination Rates for Children with Chronic Conditions Kevin Dombkowski, DrPH, MS April 20, 2010
Guiding Questions How accurate is the MCIR high risk indicator? How effective is this indicator for improving flu vaccination rates among children with chronic conditions?
MCIR High Risk Indicator Accuracy of MCIR high risk indicator was assessed via parent telephone interviews: physician diagnosis of asthma asthma symptoms, medications and health services use activity limitations Interviews conducted: March – June, 2008 February – May, 2009
MCIR High Risk Indicator Overall, 89% confirmed asthma or similar condition: asthma: 83% wheezy cough: 5% reactive airway disease (RAD): 1%
Objective To assess the effectiveness of seasonal influenza vaccination reminder notices for children with high risk conditions using a statewide IIS
Methods Used the Michigan Care Improvement Registry (MCIR) Pilot tested reminders in 3 local health departments: Kent Kalamazoo Ottawa Identified 3,618 high risk children 24-59 months (October, 2008)
Methods Randomized into two groups: Reminders: Mailed reminders No reminders Reminders: mailed November, 2008 sent via First Class USPS mail, marked “return service requested”
Methods Determined validity / accuracy of mailing addresses using the US Postal Service National Change of Address (NCOA) process Classified as undeliverable: Invalid addresses COA >1 year prior to notification date COA identifies out-of-state move
Methods Assessed flu vaccination: Primary outcomes assessed: Oct. 2008 – Feb. 2009 Primary outcomes assessed: Flu vaccination status Time until flu vaccination Used MCIR and Medicaid claims Assessed bivariate and multivariate associations
Study Sample
Flu Vaccination Results
Flu Vaccination Child Characteristics n (1,946) Flu Vaccine % Age (months) 24-35 630 28 36-47 669 26 48-59 647 Crude association between age and shot after notification using logistic regression: 24-35 mo: Reference 36-47 mo: OR = 0.9, p = 0.8 48-59 mo: OR= 9.3, p = 0.7
Flu Vaccination Child Characteristics n (1,946) Flu Vaccine % Race/Ethnicity White 975 25 Black 497 24 Hispanic 373 34 All others 101 26 Crude association of race and shot after notification (Y/N) using logistic regression: White = Reference Black OR = 1.0, p = 0.1 Hispanic OR = 1.6, p = 0.001 Other OR = 1.0, p = 0.7
Flu Vaccination Child Characteristics n (1,946) Flu Vaccine % Flu Vaccine in Previous Season Yes 1223 30 No 723 21 Crude association between flu vaccination in previous season (y/n) and shot after notification (y/n) using logistic regression: Yes: OR =1.67, p=<0.0001 No: Reference
Flu Vaccination Child Characteristics n (1,946) Flu Vaccine % Medicaid Enrollment in October 2008 Yes 1711 28 No 235 17 Crude association between Medicaid enrollee in October 2008 (y/n) and shot after notification (y/n) using logistic regression: Yes: OR =1.94, p=<0.0001 No: Reference
Cumulative Immunization (n=2552)
Cumulative Immunization (n=2552)
Cumulative Immunization (n=2552)
Cumulative Immunization (n=2552)
Cumulative Immunization (n=2552)
Cumulative Immunization (n=2552)
Notification Mailed notification was associated with earlier immunization, controlling for: age race/ethnicity history of flu vaccination Medicaid enrollment Undeliverable notices have a negative impact on flu vaccination rates Multivariate modeling methods in SAS: *TPHREG *Cox-proportional hazards *Modeled time until 1st shot, censored shot after notification (y/n) Notification: HR = 1.22, p = 0.03 Influenza Vaccination in Any Previous Year: HR =1.54, p =<0.0001 Medicaid Enrollment in October 2008 : HR=1.82, p=0.0003 Race/ethnicity: Black: HR =0.94, p= 0.58 Race/ethnicity: Hispanic: HR=1.39, p=0.003 Race/ethnicity: Other or Unknown: HR=1.12, p=0.58
Impact of Undeliverable Notices
29% 24%
30% 23%
Limitations Pilot implementation in 3 LHD jurisdictions Implementation varied between LHDs, including format of message
Conclusions It is feasible to target reminders for seasonal flu vaccination using a statewide IIS Seasonal flu vaccination rates among children with high risk conditions were substantially higher than historical rates
Conclusions Reminder notices from the local health department were associated with a modest increase in seasonal flu vaccination Improved reminder deliverability could be achieved by address corrections prior to mailing
Implications Targeted reminder notices to children with chronic conditions may be effective mechanism to increase seasonal flu vaccination rates Improved parent contact information may lead to increased effectiveness of seasonal flu vaccination reminder programs
What if a Shortage Occurs? May be especially important to identify priority cases when flu vaccine supplies are limited
Acknowledgements Many thanks to my collaborators at: University of Michigan Michigan Dept. of Community Health Michigan Care Improvement Registry Kalamazoo, Kent and Ottawa County Health Departments Centers for Disease Control and Prevention
Thank you for your attention! kjd@med.umich.edu