02/16/2007 Meeting Minutes

NERHMIS Meeting Minutes

February 16, 2007

Participants:

NH- Bob Sparks, Linda Newell, Chris Pitcher
MA- Bill Silvestri, Jonathan Sherwood
ME- Cindy Namer
VT- Brian Smith, Richard Rankin
TA providers- Kat Freeman, Michelle Hayes, John McGah
HUD- Maryann Martel, Roslyn Block

Joined via conference call

MA- Matt Simmons
CT- Ken Teele, Natalie Mathews

Point in Time De-Brief- Lessons Learned

Massachusetts

Brockton - first attempt not very successful
Quincy (6 providers in CoC)- full participation with all agencies; used excel spreadsheets to track people and audit against SHORE HMIS; filled out on paper and then entered into excel spreadsheet; numbers tabulated automatically, created unique ID from identifiers found 8 duplicates
Data- identifying data except for DV shelter only collected pops/sub pops data from Chart K without identifying data (high percentage found to be victims of DV)
Counted 34 people on the street

Challenged by collecting data for those on the street (either mental health or substance abuse issues)
Hard time assessing chronic status for those on the street especially when the person is unwilling to provide information
May be able to obtain gender, age decade, etc.
Biggest challenge- defining chronicity on someone that is not known by providers

Numbers are expected to be collected for state by March 5^th

Vermont

Last year:

Bennington 233 unduplicated (.062 % of general population)
Less than 1000 people counted

This year

2000+ counted for the state during PIT using single form
Television coverage

New Hampshire

HMIS staff worked with Manchester- modified VT tool

Poor participation, several agencies chose not to participate even though they had training and tool
Count of 25^th and 26^th- published data on the 13^th
Used survey monkey for analysis (not entered into HMIS)
Counts: Sheltered count, Service counts, Street Count (2-3:00 AM)
Numbers from school system (includes doubled up count) from McKinney Vento homeless coordinator
Duplicated count 726- includes 100+ doubled up-- (unduplicated verifiable account- 360) actual number is somewhere between the two (actual bed capacity is 420)
Last year reported 1200 (believe is very high)

Bob worked on BOS
Next year hope to conduct on statewide PIT

Maine

Awaiting receipt of numbers
Portland 78% participations (high coverage for family, ind, youth- lower participation in homeless camps)

Paid leaders of the camps to get participation

BOS and Greater Penabscot- compiling numbers by March 15; hope to publish numbers by beginning of April

Sent out 670+ forms (received majority back)
MHSA doing data entry into custom database (Portland will also enter into same database)
Full case management assessment with income, health, use of benefits, etc. (no identifiers entered into separate database)

Has training protocol

DC (Richard shared from working with them)

Approach- look at everyone that is in the shelter on the 25^th
TCP verifying data with each provider
Data was entered into HMIS

Verified data for clients were exited appropriately before the day of the 25^th
High levels of participation
Shelters and agencies have to certify the bed list is complete

Conference Planning- Review of Draft Agenda and Identification of Speakers

See separate conference handout from John McGah

PIT Count Session
Theme: Using the Data
Decision- minimal paper with a CD with content
Capacity: 200 persons

Bay Area Communities Homeless Information Collaborative (BACHIC)- Discussion on process of aggregating regional data- lessons from the Bay Area (Ray Allen, Steve Sangiovase, and John)

11 CoC’s in Bay Area (basically county level CoC)
Geographic area is about the size of MD (pop about 7.5 million)
Schwab foundation originally founded the group who were looking at addressing homelessness regionally

Foundation paid for initial facilitation costs and later planning stages of the data warehouse

Each CoC is represented by a coordinator
Each CoC has their own HMIS (not all the same vendor product) but was interested in understanding mobility patterns across CoCs
CTA- local non-profit dealing with technology to give advise to the group on HMIS systems (has previously worked with 7 other CoCs with different foundation funding)- brought and advised them data warehouse was a best approach as they did not want to share data across CoC
Group decided they wanted to develop a data warehouse (RFP to Stanford, Berkeley, and CTA)- CTA won the bid
Foundation put up initial $ to develop the warehouse ($100,000)- foundation pulled out of funding homeless issues
Approached HUD to fund as a pilot best practice project- agreed to provide money to develop warehouse as a best practice ($250,000 grant)- documenting experiences (planning, governance) and sharing models of developing data warehouses (how did they build the warehouse- technical steps); policies and procedures on building a data warehouse and the agreements put in place to ensure buy in for the steps for the process
Challenges have not been with the technology but with politics of getting buy in from the CoCs
5 different softwares (6 CoCs using the same system) and various methods of who was hosting local data
Data warehouse was the only avenue open to aggregate information
Universal and Program Level Data (w/out Name, DOB, and SSN)- would rather submit

Needed to have zip code to track movement
Year of Birth to track demographics by age
(took a few months to reach agreement)

Using CSV to export data out of each system (and not XML)- a number of CoCs did not have local expertise to put data in XML
Data standards minimally requires in CSV export value (prior to publish of the HUD CSV standard which seems to be far more complicated than they expected)
Hoping over the life of the project to move to XML
Data will be collected in standard form and build the warehouse around that
XML specification structurally requires client data to be within a program (CSV spec does not make that a requirement)- initial challenges: budget and commitment of level of effort at the local level
Balancing act between what they can request and what they can require
Generate snapshot of data on monthly basis- will only receive client level data accurate at the end of the month and not on all changes that take place during the month (more general based on date ranges)
Unique algorithm- taking components of first name, last name, gender, and year of birth information
Regional unique identifiers within the CoC and then de-duplicated across CoCs
CoCs were worried about re-identification from RHINO administrators at CTA
Tested 3 CoC databases for a clash rate to identify if the algorithm sufficiently identified duplicates (3.3% duplication on top of what SP naturally did)
Many of the CoCs do not have funding for HMIS- CTA is trying to minimize work for local system administrators
What kind of funding are they going to have to maintain the system in the future?

HUGE problem- the group are having monthly conversations about how they sustain project after Dec of this year (when last of CoCs dumps their data into it)
They will have the data- but need to raise additional funds to analyze the data- may approach local universities to conduct research

Data belongs to each of the CoCs and nothing can be done with the data without their approval (not MOU have guiding principals)

Group took a vote that each CoC agreed to abide by the specifications in the guiding principals

Maine

Moving toward XML data export
Now paper reports for ESG and HMIS reports collective (14 days to submit matching reports)

Next Step:

Cindy share MHSA PIT training module
Michelle talk to HUD re: aggregating PIT counts for the region

March 16^th- Meeting

Follow up with Barb Ritter to join March meeting re: approach to release of data
Conference- dedicated state contacts need to participate- need to bring names, titles, and email address of ICH and state policy academy
Letter re: funding HMIS outside of CoC
Eric Hirsch for DV discussion- for next discussion
Richard identify provider for DV data entry into HMIS