Correspondence between

Okay, thanks, Bob. Hf incorporate the previous years and put the new results into a publisher document with some description of what’s going on in the chart. We'll find out ff this changes the results!

Cristin, toan safely say | don’t understand it without same explanation. However, late yesterday Nancy and | decided to take a look two years farther back to see how they fit with the recent years. Please see attached, which i just finished this marning. Sorry / didn’t have this for you yesterday before you did the analysis. Would this new information change the analysis? 2 xxxEND_PAGE:dhhs03_b0833_3004_3007_2 003006 Thariks, Bob

Nancy and Bob, Is the control chart clear enough, do you think? | could whip up a fact sheet with a description of what it shows, if you think it would help make it more digestible for our audience. Cristin

Nancy and Bob, is the corrtral chart clear enough, do you think? | could whip up a fact sheet with a description of what it shows, if you think it would help make it more digestible for our audience, Cristin

Hi Nancy, imiade a p-chart, which Shewhart’s version of a control chart for nronortions, for the data you sent. Basically, | used the monthly data from 2015-14 to create upper and lower control limits, then plotted the 2014-15 data in arun chart. it shows that the three months in question are the only onas that lie outside the control lirnit: in fect, they are the only points that fle well above the meari at all. This doesn’t say anything about causality, but it does warrant further investigation. There are several next steps we can ernploy if the folks upstairs ask us to look deener inte the data. Alsa, Yr not sure if you talked at all with the Environmanial Health folks, but their tanicologists could probably heto give us sarne context to the issue. Cristin

xxxEND_PAGE:dhhs02_b0237_0797_0885_16 000813 Okay, thanks, Bob. I'll incorporate the previous years and put the new results into a publisher document with some description of what's going on in the chart. We'll find out if this changes the results!

Thanks, Patti! It took me a while te get it right, but i was able to create a p-c j version of a contral chart for proportions. I’m nat sure if you can read the attachrnent, out | us ly data fram 2013-14 to calculate a mean proportion, mean quoiber of kids tested per month, plus upper and law ilienits. the 2614-15 data on a p-chart with the mean, LCL, and UCL cale previous year. The chart shows that the UCL for the year. ideally, we would want to use more years of data to establish the control limits. Also, mapping the cases might heip figure out c pipes were the source of exposure. Sarah also suggested a Poisson regression (factoring in such infermation as when the é }, if the folks if reporting changed at all dk uring the exposure period (e.g. were there more capillary tests or tests ed the mont contr above Fit looks lay ta you, I'll send it off to Nancy as 500n as | get in tomarrow. (She cali over here this afternoon asking for it, so rm sure she’s anxious to send it upstairs.) Good night, Ps