Missing Assessment Data? Well, define "missing"...

Share on:

Over the past couple of years, we have come to realize that between the CoC APR, the HMIS APR, the AHAR, and the NOFA, the definition of "missing" assessment data is a moving target. An agency with a completely clean APR that shows no missing data may have many many errors contributing to bad data on the NOFA or the HMIS APR. This realization brought much frustration for our users who had been using their APR to tell them if they had any missing data. The APR happens to be pretty lenient as to what it will accept as complete data compared to the other required reports. So when it was time to submit the HMIS APR, the NOFA, and the AHAR to HUD and we were calling up users to fix lists of clients with missing data, some of them were a little peeved because they had been working diligently to keep their APRs looking good, and now we were heaping on more mistakes to fix. They wanted to know, "How do we know for sure that we're really all good?" One problem here is that they were using the APR as their data quality report. Bowman does recommend that users use the APR for data quality, but I suggest that it is necessary to discourage users from using the APR as their sole means of checking data quality.

If users are not to use the APR for their Data Quality, what should they use? The 0252 Data Quality Report Card is an option, but it is also too lenient when it comes to catching missing data. Like the APR, it does not catch situations where an assessment question was answered, but not backdated correctly. These two reports will not bring your data to perfection, as missing data goes.

A challenge that we had was figuring out what each report's definition of missing is and then landing on a decision about which report we were going to use for assessment completeness. And I say "assessment completeness" instead of "assessment data quality" because there is more to data quality than simply having all the data elements saved. Either way, to measure completeness, we wanted the report with the most stringent definition of "missing" as possible so that our users could use that to stay on top of their missing data.

The report we landed on was the HMIS APR. It picks up not only assessment data that has not been backdated correctly, but it also picks up clients with open services, which, while that may not be important for some implementations, it is for ours.

A second problem is now that we know which HUD report uses the most stringent definition of missing, users cannot successfully run the HMIS APR as it is written. The prompts are clearly meant for HMIS admins to answer, and deciphering the Detail tab relies on Input Controls. (See my post called Why I Snark on Input Controls for more on this.)

This is another blog post, but the way I handled this was I modified the HMIS APR, adding in a tab similar to the image below.

Columns: Client ID, Missing Name, SSN, DOB, etc. Modified HMIS APR

I took out all of the tabs that would be irrelevant to users and set up alerters to show "DKR" if it was answered "Don't Know" or "Refused", and "Missing" if it was missing according to the HMIS APR.

I can explain more about the modifications I made in a another post, but the point here is that, short of a comprehensive data quality report from Bowman, it is best to use either a custom report you design or modify the HMIS APR for your users to get a real picture of the completeness of their assessment data.