AHLTA

1. AHLTA Data Quality Management Course 10 February 2009 DHIMS Sustainment

2. Known Issues • New Issues • Error Code 102 False Positives • Update on Old Issues • Duplicate Patients • ADM Writeback Errors • Signed encounters that have not synched to the CDR • Inferred/missing SADRs

3. Why Duplicate Patient Records • Unintended human error during patient registration • Typographic errors • Critical errors (EDI_PN, SSN, Last_Name) • Less critical errors (MIDDLE_Name, Gender, SPONSOR_SSN, etc.) • Inaccurate information from source • Workflow-induced data entry issues during patient registration • Time pressure • Multiple eligibilities • Pseudo-SSN • Automated errors • Person Identification Service (PIDS) “fetches”, etc.

4. Attention to detail at the MTF • Ongoing assessment of error generation mechanism • No new items to report (see next slide) • Manual process developed to correct errors • Teamwork effort of site and TMCI contract team • Currently on hold due to drug-allergy information cancellations • Fix to arrive for testing 2/13, estimate resume merges 2/20 • Automated error correction • Contract awarded in December, automatic merges should begin in March (assuming above fix is in place) • Clerical staff issues • Air Force study; clerical staff training on duplicate record causes • Error rate dropped from 25% to 15% • Limit number of people with access to registration

5. Major causes of duplicate patient records • If accessing the master patient index is slow, users will often create a record "to get the patient seen"; this produces a duplicate record. Commonly, though the wait is not usually long. • DEERS has a very rigid trait matching system, so a mis-type will fail to find the patient's record. If the user doesn't review his/her input to detect the error , he/she might create a new record, again, to "get the patient seen". • We are looking at purchasing a modern trait-matching service that will cut way down on this (time-line not determined yet) • If there is a significant enough difference between the patient's DEERS demographics and the data that is entered, DEERS may recognize the person as a new patient. • The prospective trait-matching service should eliminate this

6. Major causes of duplicate patient records (cont’d) • If the patient gives incorrect demographic information to the user (error, language barrier, nicknames, non-western-type names), the same will happen as noted in #3 on the previous slide. • Use of Pseudo-SSN's (will almost always result in duplicate record): • Not everyone whose care is documented in AHLTA has a SSN yet (e.g. newborns and foreign nationals), but DEERS requires one • The necessary workaround is to create a "pseudo-SSN", to be replaced with the correct one when one is issued • There are rules about what numbers to place where in order to show that it is a PSSN rather than a true one, but sometimes errors occur • Also, the MTF has to keep track of these patients in order to substitute the real number when it is available--doesn't always happen

7. AHLTA and Patient Merge… • Current manual process for merging duplicate records • Duplicate records are detected either by provider at point of care or by automated “comb” of MTF LCS; MTF creates Trouble Ticket(s) • TMCI (Tier 3) Triage group receives and manages all MHS trouble tickets • After investigation and verification that it is truly a merge issue the information is sent to DISA • DISA then combines the different records into one “KEPT” record • Automated identification and resolution • Automated system will find duplicates and resolve them • Current rate of duplicate record creation is about 18000/year • There are currently 1.3 million duplicates identified in CDR – automated system will clean up

8. Error Code 102 Exceptions • ADM Processing Error - Failed WAM Validation • Invalid or Out-of-Sync ICD9 codes and CPT codes • 99371 is not a valid code value (99372,99373,99361,99362) • In addition have “False Positive” 102 errors (record appears in error report, but data is correct) during Oct/Nov 2008. • Error language is "at least 1 ICD9 code must be present" • This is fixed day forward but backlog is just now being cleared; No spreadsheet yet • There was a significant increase in EC 102 failures since March due to obsolete E&M codes being removed from the table without the replacement E&M codes being available for selection in AHLTA • These can be manually coded in CHCS by the site but they will still show on the AHTLA ADM exception report • We can have the site run an ad-hoc report in CHCS to confirm which failures were coded manually and change the status in the CDR so it will no longer be reflected as an EC 102 exception • This was remedied day-forward when all sites upgraded to 838.23. Fix of backlog is ongoing through NG Tier III • Backlog at all sites completed except: • Five sites have not sent in required information after 2 requests • One is still in process • 76% of the Oct EC 102 failures enterprise-wide are false positives failing with error: “At least 1 ICD9 code must be present”. This has been fixed day forward with the 3.3.3.1 mid-tier update 1 (3.3.3.2 MT) for all sites. NG will now work with sites to reset to 120. Encounters also need to be synched.

9. Error Code 103 Exceptions ADM Processing Error - Undefined CHCS Error Types of EC 103 errors Disposition is required Invalid Modifier MOD1 for EM code Injury Cause Code is missing M/Objects errors Enterprise Wide Problems (Fix is re-queue done centrally by NG) “Disposition is required”: Prior to the setting of the ADMp1p2 variable, CHCS/ADM was accepting the TCONS without dispositions (which is correct—the patient is not there). Afterward, the system required a disposition, with no method of entering one. Fix was implemented in November 2008 17 Sites have not sent in required information (after 2 requests) to NG for re-queue There is also a TCON error where there is no IEN Fix was implemented in November 2008 19 sites have not sent in required information after 2 requests Two sites await MObj 97 installation One site has cleanup in process

10. Signed encounters that have not synched to the CDR • These encounters are not seen on the ADM exception report even though there have been completed in AHLTA but not in CHCS • Problem fixed day-forward in Nov 08 but historical cleanup just now getting started; No progress spreadsheet available yet

11. Inferred/missing SADRs • Began increasing in January 2008 due to mismatch between FY 07 and FY 08 CPT codes • Steady decrease toward baseline since June 2008 due to implementation of the FY 08 CPT codes and NG re-queuing of the encounters

12. Questions???