Data Quality in the State of New Hampshire Donna Curley – HMIS Project Manager Al Vitale - Data Analyst Bowman Eastern User Summit July 19-20, 2011
Why Data Quality? HMIS is HUD’s response to a Congressional Directive to capture better data on homelessness Regular measurement of data quality and outcomes related to critical processes as well as mandated HUD reporting Identify issues such as training needs
Why Data Quality? Essential for community partners planning Understanding of the nature and scope of homelessness Funding applications Public awareness and education
Why Data Quality? Every CoC is required to implement an HMIS Every CoC is scored on this obligation as part of the annual CoC Competition National data on homelessness is vital for HUD reporting and informs key policy decisions
Data Quality Framework Client Record Definition Ensure HUD mandated data is captured Accuracy Data must be entered correctly Timeliness Data entered promptly for reporting requirements
Data Quality Framework Completeness Information is correctly entered on all clients Consistency Performance information is consistent across time Incentives and enforcement Public recognition for compliance Funding impacts for non-compliance
Data Quality Factors Prioritized process in the Agency Give staff the time to participate in training and to complete entry The environment/equipment arranged to support entry Is the data used?
Data Quality Factors Trust Client honesty is based on establishing rapport and the quality and content of the interview process. Staff to System Staff will elect not to enter information if they don’t trust. Client to Staff Clients won’t tell the truth if they don’t trust.
Common Data Quality Errors Issues with Training Entering “no” when you mean “yes” Failing to enter information on some Clients Date Errors (DOB is 7/11/1972, entered 7/11/2011)
Common Data Quality Errors Transposing numbers Spelling errors in names such as Glen vs. Glenn Accidentally selecting the wrong response from a drop down
Data Quality Training New User Site Administrators Policies and Procedures Data elements/definitions/entry guide Help Desk/Technical assistance
Ensuring Data Quality Provide monthly reports to agencies for review Clear direction for correcting data Procedures for correcting are defined Staff to look at data reliability and validity issues Adjust & add data quality reports
The 3 data element categories Program Descriptor Data Elements (PDDE) Universal Data Elements (UDE) Program-Specific Data Elements (PSDE) The Data Standards define specific, acceptable responses for each data element
Incongruity (Wacky Data) Reports Entry Date Precedes Birth Date Client records are displayed if s/he is entered into a program before being born. Pregnant Female Client records are displayed if she is pregnant and less then age 15 or more then age 55. Pregnant Male This report displays client data if he is pregnant. Questionable Age Client data is displayed if s/he is less then age 0 or more then age 100.
Incongruity (Wacky Data) Reports Age and Disabilities If a client is less then 18 years of age and has Alcohol or Drug Abuse disability types the record is displayed. Age and Head of Household This report displays client records if the client is designated as a “Head of Household” AND is less then 18 years of age. Age and Veteran If a client is designated as a Veteran and the client’s age is less then 17 the record is displayed. Children Not in a Household This program reports client data if the person is under age 18 and is not affiliated with a household.
New Hampshire BOS Data Quality How are we doing? Before Data Quality (1/1/05 – 6/30/08) Number of errors per Client 3.26 (of 4414 clients) How are we doing? After Data Quality (7/1/08 – 7/1/11) Number of errors per Client .62 (of 6231 clients) A significant improvement!
Lessons Learned Partner with Agencies, buy-in is crucial Establish Goals such as 100% AHAR involvement Identify mandated elements and be clear on the definitions Ensure data entry requirements are attainable Ensure agencies are accountable Have a plan for training and assistance
Data Quality Reports Use them to understand if data is entered timely, completely and accurately Generate them regularly (monthly or more frequently) Have a process to allow for data correction We have an obligation to remember that every HMIS record is a survivor story so we must understand the client need, and to do that we must collect accurate data
Donna Curley, HMIS Project Manager d.curley@harborhomes.org Thank you! Donna Curley, HMIS Project Manager d.curley@harborhomes.org Al Vitale, Data Analyst Alfred.Vitale@dhhs.state.nh.us http://nh-hmis.org