Arizona Behavioral Risk Factor Surveillance Systems Annual Meeting
Arizona Behavioral Risk Factor Surveillance Systems Annual Meeting Judy Bass Arizona BRFSS Coordinator August 23, 2016
Speakers Robert Bailey, Bureau Chief, Public Health Statistics Arizona’s BRFSS Project Director Samantha Vincent, ICF International Research Associate
BRFSS DATA RELEASE – Change in Policy Data Use Agreements implemented to protect data Ø Paris School of Economics’ Publication “Where the Streets Have A Name” ü Used Arizona BRFSS Data ü “Re-identified” ü BRFSS respondents to the neighborhood level (not individual level) by linking to other public data sources
Highlights the risks of re-identification from “Big Data” Ø All BRFSS public data sets modified to remove - potentially identifiable fields (county, zip code) Ø Public Data sets now comply with the Federal “Safe Harbor” - standard for de-identification. Ø Limited Data sets (including those two variables) - require a restrictive data sharing agreement - prohibiting any attempt at linkage or re-identification
BRFSS SURVEY Changes in Administration Ø Annual 2013 Report http: //pub. azdhs. gov/e-books/brfss-annualreport-2013/index. html (Also available by e-book) Ø Annual 2014 Report will be published by 9/30/16 (Also by e-book)
In 2015, the Arizona BRFSS Accomplishments Ø Ø Ø Implemented a New BRFSS Website display - American Sign Language video - BRFSS “Public Use Data Files” are updated in SAS & Excel format from Survey Years [2000 – 2015] Changed the Exclusive Caller ID line to link to ICF Added a link to opt out of the survey as well as an Exclusive Caller ID Line was added to address respondent’s questions or concerns.
Arizona’s BRFSS 2016 Goals Ø Arizona Survey Sample Size is currently at 10, 500 Ø CDC requested TWO changes to Arizona’s 2016 survey: ü Changes in Family Planning Module - Now surveying women up to 50 years of age (instead of 45) to supplement national research on sexual transmission of Zika virus. AND ü Implemented sleep disorder questions in Arizona’s survey. Ø Currently working on increasing Marketing efforts for BRFSS ü First publication on the Directors’ Blog posted -July 20, 2016
Arizona’s BRFSS 2017 Goals Ø Participate in CDC’s request for oversample Americans Indians and Alaskan Natives (AI/AN) - Arizona BRFSS will include 530 completed surveys among these groups - In the past Arizona BRFSS’ surveys collected over a three-year average sample size of American Indians /Alaskan Natives Ø Collect 7, 500 interviews Ø Sample Design of 10 regions
Arizona BRFSS Survey Comparisons by Region & Completed Interviews [ 2011 -2017]
BRFSS SURVEY HISTORY – Changes in Methodology In the 2011 BRFSS weighting method changed: Ø Post-stratification to Iterative proportional fitting to include: ü cell phone interviews ü selection of units with unequal probabilities ü Increased coverage of the Arizona’s population ü data representation from local areas Remember – you can not do comparable analysis with data prior to 2011 BRFSS
Arizona BRFSS Small Area Estimation Feasibility Project Ø Uses 2014 -2015 data because they used raking, and asked nearest intersection question Ø Seeks to produce sub-county level estimates for Maricopa and Pima Counties using the raking method, based upon PCA geographic assignment Ø Will evaluate those estimates for such factors as: § Accuracy/reliability of result § Complexity of methodology § Cost of methodology § Alternative methodologies that might mitigate problems with the raking approach
Nearest Intersection Question Nearest intersection will continue to be asked (2015, 2016, 2017) at least until the feasibility of small area estimation using BRFSS has been completed to retain their potential use for Small Area Estimation ü Because of their potential for re-identification, the nearest intersection data elements cannot be made available for release to protect the anonymity of BRFSS ü Initial results are expected this fall, with a full report by the end of 2016 or early 2017
CDC & AZDHS BRFSS & OTHER HEALTHCARE STAKEHOLDERS $747 091 $800 000 $589 059 $638 348 $661 236 $619 665 $643 020 $700 000 $600 000 $500 000 $400 000 $300 000 $200 000 $100 000 $0 2011 2012 2013 2014 Year 2015 2016
How much time does it take to complete an interview? Ø How much time are respondents willing to spend in completing the BRFSS survey? Ø How are we doing in comparison to the other states?
BRFSS 2016 Interview Length of Time 30 25 24, 4 25, 4 27, 8 AVG 24. 15 Min. 24, 7 21, 2 23, 5 25, 3 20, 9 Minutes 20 15 10 5 0 Arizona Arkansas Connecticut Maryland New Hampshire Rhode Island Vermont Washington
Minutes
Core Questions 120 107 98 Number of Questions 100 97 89 85 93 90 80 60 40 2011 2012 2013 2014 Year 2015 2016 2017
Optional Module Question 50 48 45 Number of Questions Asked 40 33 35 29 28 30 25 25 19 20 11 15 10 5 0 2011 2012 2013 2014 Year 2015 2016 2017
State-Added Questions 58 60 Number of Questions Asked 50 45 50 42 40 40 29 30 16 20 10 0 2011 2012 2013 2014 Year 2015 2016 2017
BRFSS 2017 Proposed Optional Module Questions # of Questions Timing Alcohol Screening & Brief Intervention 5 0. 44 Diabetes 9 0. 29 Pre-Diabetes 2 0. 25 Preconception/ Family Planning 7 0. 92 Sugar Sweetened Beverages 2 0. 65 Total time 25 2. 55
BRFSS 2017 Voting Ballot Type of question CORE QUESTIONS Total Number of questions Total Number of minutes 90 16. 87 Optional Modules Preconception/Family Planning Sugar Sweetened Bev Pre-diabetes Diabetes Alcohol Screening 7 2 2 9 5 . 92. 65. 29. 44 State-Added Food Assistance Nearest Cross Streets Access to Care Substance Abuse Prescription Drug-Abuse Hearing Aid Children with special needs 3 2 3 1 2 . 31. 93. 56. 36. 44. 22. 20 131 22. 44 Grand Total
BRFSS 2017 – Survey is Currently Timed at 22. 44 Minutes. Ø In the 2014 and 2015 a split survey was necessary or order to stay within an average length of less than 28 minutes. The effect that using a split survey has on data analysis really depends on the sample size. Ø The split is highly recommended to obtain a better representation of the data collected. CDC does require a minimum of 2, 500 per split, but keeping in mind 500 per Stratum/Regions. Ø Ten strata/ regions is recommended for the Arizona BRFSS 2017.
Cost per questions & Cell / Landline Ratios to Change Ø State-added and Optional Module costs will be $7, 100 per question for 2017 BRFSS Survey Ø CDC is planning to increase the ratio of cell phone interviews from 35% to nearly 60% for the BRFSS 2017. Ø Landline ratio will reduced from 70 % to 40% Ø The cost per question for 2018 Survey will remain at $7, 100 per question.
Voting Process �Those applicants who are present will now be asked to vote on the 2017 questions
BRFSS 2017 Survey Polling
State-Added / Module Voting • Vote Yes for question modules that AZ should include in the BRFSS 2017 survey • Vote No for question modules that AZ should not include in the BRFSS 2017 survey
Should AZ increase the current sample size of 7, 500 an additional 530 units to support the oversampling of American Indians and Alaska Natives? 1. Yes 2. No
Should AZ adopt ALL State-Added / Modules provided on the hand-out? 1. Yes 2. No
Distribution of Questions State-Added 40 48 45 33 98 85 2011 2012 42 Optional 58 Core 29 25 11 29 2013 50 19 28 107 16 89 93 2015 2016 97 2014 90 2017
Thank You!
- Slides: 31