Known data errors

Some records have known errors or data that triggers false positives in consistency checks. L Problem is a record of these so they can be excluded from queries. The table would usually be populated by the data collector during vetting.

Instructions for adding a record

If a false positive or known uncorrected error is found:

1. open table L Problem from Centralized data front end.accdb
2. make a new line
   • D_ID field: the D_ID of the record
   • Query field: the name of the query that detected this record and from which it should be masked
   • Notes field: explanations, eg why the error was not corrected or why it is a false positive

Instructions for using the table in queries

For every query, filter the L Problem table to only records related to that query and only list if null.

List of queries that use L_Problem table (SMW)

(Based on Property:DICLProblem as set in Template:Data Integrity Check.)

	Status
Query NDC dx implying death across encounters
Link_suspect_visitAdmitDtTm_mult_to-from-home query
Link suspect mismatch pre inpt ours incomplete query
Link suspect mismatch pre inpt should be ours incomplete query
Query cardiac arrest throughout admission
Link suspect negative transit time query
Link suspect dead then alive query

Location of List of old known errors

The Data processor keeps an Excel workbook that includes: reason some profiles are missing data elements (APACHE - 8 ICU profiles), gaps in database because of deletions including reasons for deletions, overlaps, overs. Example missing all APACHE elements in TMSX, MedTMS and reason why could never be obtained.18:01, 2014 May 2 (CDT) M:\GAPS_OVERLAP_OVERS

to do

• need to discuss linking queries with Pagasa when she is back from vacation Jul 21.
• Discussion about 8 profiles missing all APACHE elements and specific reason why missing etc.