Assessing Change with IIS Steve Robison Oregon Immunization Program
Assessing Change with IIS Steve Robison Oregon Immunization Program
Overview IIS Use in Public Health Static vs Dynamic Assessment Working Around IIS Imperfections How to Use IIS Strengths Modeling Change Examples of IIS Use for Change
IIS & Immunization Surveillance IIS started as clinical support tools In late 1980 s, IIS role expanded to surveillance Goals of early surveillance were population rate assessment Looking for pockets of need Generally done annually- not on the fly long lag times for reporting and processing
Two Types IIS Immunization surveillance has two types: Static (rate) assessment Done at regular intervals Predictable, repeatable, routinized Useful for tracking trends Long period of analysis- (ie: rates per year) Dynamic assessment Done on the fly Typically question-driven Shorter time-interval focus and quicker reporting Likely customized for situation; may not be repeated Dynamic Static
Type Examples Static Yearly two year old rates Yearly adolescent rates Dynamic Hep A uptake in a local area after a public health announcement regarding a food-borne outbreak Week by week influenza immunization uptake during a season
Dynamic Impetus: Did Something Change? Question-based in the form of: did X change Y? Typically follows something in the real world: Either something we did (example: outbreak response) Or external, real world event (example: news story about deaths from VPD) Specific, affected population, geography, or time.
Working With IIS Imperfections Issues of capture/accuracy/mobility can weigh heavily against generating static rates; But such issues may not matter if our question is about change. Basic principle- looking at change as likelihoods(ratio of rates) will cancel out most IIS data imperfections and biases.
Likelihood Example For HPV series completion- Assume we have 140% IIS denominator inflation, and 10% under-reported shots for 13-17 year olds; If a true HPV UTD rate is 60%, the IIS rate will be 38.6%. If a true increase of 10% occurs in the next year, then the IIS increase will appear to be 6.4%. Ratio of 70%/60% = 1.17; Ratio of (38.6%+6.4%)/38.6% = 1.17. IIS ratios of change are more likely to be accurate than simple rates
Jan1 Jan7 Jan13 Jan19 Jan25 Jan31 Feb6 Feb12 Feb18 Feb24 Mar2 Mar8 Mar14 Mar20 Mar26 Apr1 Apr7 Apr13 Apr19 Apr25 May1 May7 May13 May19 May25 May31 Jun6 Jun12 Jun18 Jun24 Jun30 July6 July12 July18 July24 July30 Aug5 Aug11 Aug17 Aug23 Aug29 Sep4 Sep10 Sep16 Sep22 Sep28 Oct4 Oct10 Oct16 Oct22 Oct28 Nov3 Nov9 Nov15 Nov21 Nov27 Dec3 Dec9 Dec15 Dec21 Dec27 IIS Strengths for Assessing Change-Stability 1400 Pneumococcal Conjugate Vaccine (PCV) Receipt at Age 65, by Administration Date, (2013-2017) 1200 1000 800 influenza component stable component 600 400 200 0
IIS Strengths for Assessing Change-Sensitive 350 300 250 Oregon HPV Immunization Counts by Day for Tweens (11-12) in 2016 and 2017, Sig. higher- what happened? First Merck ad period 200 150 100 50 0 Age 11-12 in 2017 age 11-12 in 2016
Modeling Change Putting these pieces together: 1.Refine base question- did something change- to did the pattern of immunizations in the IIS change in a short time period? 2.Two flavors of this- did an immunization total jump after an event? Or (more sophisticated) did an increase above seasonal pattern occur? 3.Calculate rates before and after events of concern- choose a short time period (week as default) 4.Calculate the ratio of rates 5.If seasonality or other base differences are suspected, repeat 3&4 for the prior year 6.Calculate the ratio of the two ratios- this is an odds ratio
Non-Commercial Product Endorsement A great little tool for these calculations is WinPepi Freeware Written by an Epidemiologist, Joe Ahbrams (deceased) Winpepi box
Example: Oregon Adult Hepatitis A Immunization After San Diego (2017) Wide media attention around San Diego homeless Hepatitis A outbreak Emergency declared September of 2017- Lifted February of 2018. So in neighboring Oregon, was there an increase in adult Hepatitis A immunization following San Diego? Oregon background rate of HepA immunization for travel; look at amount above background for effect.
Number of HepA Immunizations per Month 6000 Oregon Adult (Age >18) Hepatitis A Immunization by Month and Year 5000 4000 3000 2000 1000 0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 2014 2015 2016 2017 2018
Oregon Adult Hepatitis A Immunizations Compared to Prior Averages and Trends, 2017-18 Compared to 3-Year Prior Means Month to Month Trend (Prior Year) Month Ratio 95% CI Odds Ratio 95% CI May 2017 0.99.94 to 1.04 0.96.90 to 1.02 June 2017 1.05 1.00 to 1.10 1.04.98 to 1.11 July 2017 0.94.89 to.98 0.93.87 to 1.0 August 2017 1.04.99 to 1.09 0.95.89 to 1.01 September 2017 1.09 1.04 to 1.14 1.16 1.09 to 1.23 October 2017 1.28 1.22 to 1.33 1.18 1.1 to 1.25 November 2017 1.11 1.06 to 1.17 0.87.81 to.93 December 2017 1.03.98 to 1.08 1.03.96 to 1.10 January 2018 1.13 1.07 to 1.17 1.11 1.03 to 1.19 February 2018 1.03.97 to 1.07 0.84.78 to.90 March 2018 1.04.98 to 1.07 0.95.89 to 1.02 April 2018 0.96.90 to.99 0.94.88 to 1.01
HAV Summary Defined the period of effect as Sept of 2017 through Jan of 2018 Compared this effect period to the prior year period as a ratio Finding that for the effect period, adult HAV increased 18% more than expected (LR = 1.18)
Remember This? Original measles case in Disneyland, CA, Jan 2015 Susceptible visitors took measles home Oregon was lucky we didn t get a secondary outbreak
Measles Immunizations in Oregon, Post- Disneyland Questions- Did the wide publicity around the Disneyland measles outbreak lead Oregonians to seek more immunizations? How long did this effect last? Methods- Look at count of measles cases per week in Oregon, prior and post announcement of outbreak.
Measles Immunizations per Week 4,500 Oregon Measles Immunization per Week, 2015 Compared to 2014 4,000 3,500 3,000 2,500 2,000 1,500 1,000 500 0 2014 2015 1 2 3 4 5 6 7 8 9 10 Week
2.50 2.00 Effects of Disneyland Measles Outbreak on Oregon Measles Immunizations, 2014 to 2015 Likelihood Ratios by Calendar Week 2.14 1.50 1.00 0.95 1.02 0.98 1.28 1.47 1.37 0.98 1.06 0.90 0.50 0.00 1 2 3 4 5 6 7 8 9 10 Week
Questions? Steve Robison Oregon Immunization Program steve.g.robison@state.or.us