Arizonas Sentinel Site Data Quality Efforts Fragmented Records

  • Slides: 22
Download presentation
Arizona’s Sentinel Site Data Quality Efforts Fragmented Records and MOGE Coding Lisa Rasmussen Arizona

Arizona’s Sentinel Site Data Quality Efforts Fragmented Records and MOGE Coding Lisa Rasmussen Arizona Department of Health Services March 30, 2011

ASIIS • ASIIS - Arizona State Immunization Information System • Arizona statute § 36

ASIIS • ASIIS - Arizona State Immunization Information System • Arizona statute § 36 -135, effective 1/1/98, requires reporting of all vaccines administered to children birth – 18 years of age • Currently, 4. 3 million patients and 45 million vaccination events in ASIIS • Provider site driven system

Reporting • ~1, 100 providers report through web application, flat file uploads, HL 7

Reporting • ~1, 100 providers report through web application, flat file uploads, HL 7 uploads, and paper forms • 57 sites still report via paper forms – 22 report on a regular basis – ASIIS contracts for data entry services • Efforts continually underway to reduce the number of paper reporting sites and convert them to web entry or other electronic reporting

Sentinel Site • Arizona’s Sentinel Site area is comprised of 7 counties in the

Sentinel Site • Arizona’s Sentinel Site area is comprised of 7 counties in the northern half of the state • Approximately 185, 000 children under age 18 • Predominantly rural with some of the most remote locations in the state

MOGE Coding

MOGE Coding

Observations • Paper reporting provider sites do not always use ASIIS. • In rural

Observations • Paper reporting provider sites do not always use ASIIS. • In rural settings, there is minimal patient movement between providers – limited provider choice. • Many paper reporting sites do not track the MOGE (Moved Or Gone Elsewhere) status of their clients.

MOGE Case Study • Provider office located in Mohave County, one of 9 provider

MOGE Case Study • Provider office located in Mohave County, one of 9 provider sites in the community that serves children • Provider office reported via paper forms for approximately 10 years • Many inactive clients still on their patient roster

Assessment 2010 • A Co-CASA assessment was done 3/3/2010 • 327 patients were assessed

Assessment 2010 • A Co-CASA assessment was done 3/3/2010 • 327 patients were assessed • 4: 3: 1: 3: 3: 1 rate was 20% for 12 -23 month olds • DTa. P drop off rate was 56%

Process • Site visit made 3/2010 • Reviewed Co-CASA results • Discussed importance/necessity of

Process • Site visit made 3/2010 • Reviewed Co-CASA results • Discussed importance/necessity of deleting old and inactive clients with office representatives • Provider site migrated from paper reporting to web entry after the visit

MOGE List • After site visit, ASIIS staff queried all active clients attached to

MOGE List • After site visit, ASIIS staff queried all active clients attached to the site • “Active” or “non-active“ determination was based on age and date of last visit • ASIIS identified ~4, 500 names of possible “nonactive” clients • The list of 4, 500 names was sent to the provider site for review/clean-up • Feedback was “less than positive”

Next Steps • ASIIS staff pared down original list to smaller, more manageable lists.

Next Steps • ASIIS staff pared down original list to smaller, more manageable lists. • Lists were faxed to the provider • ASIIS offered to code the records for the site • Lists of children <36 months of age were targeted first.

Lists Sent to Provider Site (in order sent) # Records Sent # MOGE’d %

Lists Sent to Provider Site (in order sent) # Records Sent # MOGE’d % MOGE’d <1 yr 44 21 48% 1 -2 yrs 172 60 35% 2 -3 yrs 299 142 47% 17+ with visit in last 6 years 365 318 87% No visit since <2004 635 560 88% 3 -4 yrs 302 130 43% 4 -5 yrs 362 172 47% Total 2179 1403 64% Age Group

Data Cleaning Results • Main MOGE reasons: – moved out of state (20%) –

Data Cleaning Results • Main MOGE reasons: – moved out of state (20%) – moved out of area (12%) – changed provider (41%) • Duplicate records were identified, corrections made, gender identified • Lists were completed in less than 3 months

Assessment Results One Year Later • • A Co-CASA assessment was done 3/11/2011 Assessment

Assessment Results One Year Later • • A Co-CASA assessment was done 3/11/2011 Assessment dates remained the same 275 patients were assessed (52 less) 4: 3: 1: 3: 3: 1 rate was 30% for 12 -23 month olds (up from 20%) • DTa. P drop off rate was 44% (down from 56%)

Lessons Learned • Smaller lists = results and cooperation • Offer to code the

Lessons Learned • Smaller lists = results and cooperation • Offer to code the records for the provider site generated good will & cooperation • Better completion rates resulted due to MOGE coding.

Fragmented Records

Fragmented Records

Fragmented Records • ASIIS has a robust deduplication algorithm, but duplicates are frequently missed

Fragmented Records • ASIIS has a robust deduplication algorithm, but duplicates are frequently missed • How to identify duplicates in 4. 3 million records?

Fragmented Records • ASIIS staff programmer created reports to identify: – Probable matches –

Fragmented Records • ASIIS staff programmer created reports to identify: – Probable matches – Possible matches • Various criteria checked: – 1 st name – Last name – Guardian name – Mother’s maiden name – Address • Date of birth ranges in manageable numbers (100’s of births rather than thousands)

New Reports 5 new reports developed – – Variation of original report Matching via

New Reports 5 new reports developed – – Variation of original report Matching via SSN Matching via Medicaid ID Baby identifiers – to match unnamed baby’s, i. e. , “Baby Boy Smith” • Mother’s maiden name or guardian name • DOB – DOB not considered as primary match criteria – still in development

Early Findings • SSN report – Remove SSNs with less than 9 characters –

Early Findings • SSN report – Remove SSNs with less than 9 characters – Remove nonsense SSNs (999 -99 -9999, 123 -45 -6789) – SSN not very good for merging duplicates but very useful for data cleaning. • Baby name report – Make sure multiple births are taken into consideration – Good for matching birth dose Hep B with other records

Lessons Learned • De-duplication is a very time consuming process • Need to automate

Lessons Learned • De-duplication is a very time consuming process • Need to automate the process to easily generate tracking reports • Work with reports most likely to match first • Limit lists to manageable sizes – “small battles instead of a large war”

Questions? Contact information: Lisa Rasmussen ASIIS Project Leader Arizona Department of Health Services (602)

Questions? Contact information: Lisa Rasmussen ASIIS Project Leader Arizona Department of Health Services (602) 364 -3619 lisa. [email protected] gov