have data, will share

I’m watching a webinar: “How Will We Know if Health Care Reform is Working?”

This webinar is offering practical advice on using data to assess the impact of the Affordable Care Act on healthcare in California. 

Really great webinar because it:

• identifies available health data resources in California (i.e. OSHPD data sets),
• describes what makes an ideal health data resource,
• suggests specific questions that could be asked of the data to learn about the impact of healthcare reform in CA, and
• gives examples of specific measures to use - knowing the measures tells a person how to manipulate the data to get answers to the questions

I felt all of it was really relevant. 

This info sesh was sponsored by the California Health Care Foundation, produced by Dr. Carmen Nevarez from the Public Health Institute, and moderated by Karen Shore from the Center for Health Improvement. The State of California’s Office of Statewide Health Planning and Development (OSHPD) was very involved and gave three presentations. Impressive meeting of the minds =)

One thing I learned: There is an organization at the University of Minnesota which helps states understand and make use of their health data assets. It’s called shadac. Really cool stuff and their publications are accessible online at www.shadac.org

Today’s webinar slides are located at http://www.dialogue4health.org/webforums/9_19_12.html#presentations

Next topics for discussion

Before I forget:

- how to plan your EMS data system to meet your business needs
- why regional data standards are insufficient to meet hospital level data needs
- hospital data systems are our information lifelines… we must set them up for success
- first responders have the first opportunity to collect high-value information upon arrival to a 911 scene… we need their data!
- legal considerations for health data exchange

how to help me open gov health data

It’s easy to find information about why we should make government data more open and readable. But it’s been a lot harder for me to find the how-to guide.

When it comes to researching “how to do things” online, I used to think of myself as pretty resourceful… but learning about how to open up gov data has been a challenge. And my program is especially difficult because the raw data I work with is personal to both patients and providers, which means it must first be rendered into a non-personal format before it can be shared.

Liability is the first thing most government people think about. This is especially true when personal health data is the topic. So I need a way to clearly explain open data and personal health data concepts to non-tech people. Here are some of the questions we’ve had:

• How do we take personal health data and make it non-personal?
• We have tons of personal health information that describes where and how people get sick/injured. This data could save lives. How can we make some of this information available without compromising privacy?
• What kinds of policies do we need to make sure our department understands what and how to share?
• Are there ever restrictions on using public data?
• Is public data the same thing as public domain?
• Do we need to publish any statements along with our public data to describe its intent and appropriate use, like private sector folks do with things like Creative Commons licensing?
• How should we keep track of what we’ve shared and with whom?

These are questions that I can’t answer on my own. But if I don’t have an action plan for getting answers, there’s no way I’ll be able to convince my colleagues or local agencies to do more sharing. Many of us have some sort of access to attorneys, but this is often quite limited.

Then there are the nuts and bolts of actual sharing. To give you tech industry people an idea of how us common folk think about this stuff, here are some highlights from my own journey:

• In the beginning, working with Excel pivot tables was a great way for someone like me (a computer lover with no programming background) to understand how computers count and sort data. This helps me set up my rows and columns in way that gives end users the most options for asking their questions and getting answers.
• Google Fusion Tables provided a very hands-on way for me to experiment with the concept of joins. Because it only lets you do one join at a time, I got to play around with the process and view the results after each step. This also made me more mindful about setting up rows and columns in published data sets.
• Tooling around on sites like GeoCommons showed me, in a very visual way, where metadata fits in to my workflow. It gave me clear examples of how to identify my data sources, describe any processing I applied, and basically take some reasonable steps to avoid looking like I stole someone’s work.

I know it’s a tall order and it’s not feasible to teach everyone how to manipulate data… but lots of us government folks need some guidelines. 

I just discovered this resource wiki at Civic Commons, which is incredible: http://wiki.civiccommons.org/Open_Data_Guidelines. I haven’t seen anything nearly this helpful to date. If anyone knows of any similar “plain English” resources for health-specific data guidelines, please let me know!!

a little EMS history trivia

Validation and widespread adoption of CPR didn’t start until the 50s. The earliest prehospital medical care systems didn’t start until around 1970. Look how far we’ve come in EMS in such short time!

Here’s the best part: We’ve only had decent Internet since the 90s. From here on out, EMS system development is sure to snowballI. With so much data and insight at our fingertips (via the web), our industry may very well be facing a renaissance period. Exciting times!

some thoughts on data quality

I’m re-reading a chapter from the Kimball Group’s Data Warehouse Lifecycle Toolkit, Second Edition (Wiley, 2008). There are some fantastic thoughts here:

Michael Hammer, in his book, Reengineering the Corporation (Collins, revised 2003) struck to the heart of the data quality problem with a brilliant observation. Paraphrasing Hammer: “Seemingly small data quality issues are, in reality, important indications of broken business processes”

I think we all agree that health data quality begins at the point of care. If our data collectors in the trenches aren’t happy and empowered to collect great data, then we’ll never get good information out of our data systems. In the medical world, I think the broken business processes are the unfair expectations we often place on our clinical data collectors. Some examples:

• Expecting collectors to understand and access multiple software programs with minimal training
• Allowing software vendors to design training curricula with little to no oversight from business users
• Expecting collectors to use Microsoft Excel to fill in gaps when software fails to deliver proper information

The good news is that I’ve never met someone in this profession who didn’t want to collect better data. In fact, many trauma registrars and ems data folks jump at any opportunity to learn how to do their jobs better. Tons and tons of passion. So in my usual tendency to exploit promising resources, I ask this question: How do we capitalize on all of this dedication and specialized talent? Some ideas:

1. Take the time to understand the data collection workflow.

Managers should do the data collection job for an hour once per month. Administrators should try it for an hour as well, at least quarterly.

2. Scrutinize the workflow and attack bottlenecks and high-risk quality issues.

Usually requires buy in from the higher-ups. 

3. Challenge traditional data management strategies.

It’s not about the software we buy. It’s about the day-to-day operations that occur around the software. These are the activities of our business users, who in my case are government workers, local agency staff, clinicians, etc. We have to fight the myth that data system compliance can be purchased from a vendor.

4. Celebrate the good people who work in the trenches.

Also from the Kimball text:

… Technical attempts to address data quality will not prevail unless they are part of an overall quality culture that must come from the top of an organization.

Data collection is hard. Recognition goes a long way. Leaders in our industry such as such as the American College of Surgeons and National Trauma Data Bank, continue to champion the trauma registry profession as technical, highly specialized, and critical to improving trauma care. Their strong example will pay dividends to our trauma systems!

silo blaster

It feels really good to be a government worker today.

I didn’t realize how much public information was out there in the world until I started working for California. Why is it, though, that I can never seem to find this “public” information when I need it? I’m pretty nosy, so if I can’t dig into something then we have a problem.

So today we are blowing up a few data silos. I assembled a few tables of emergency medical services data (all of which is already considered public) and put it all in one place. Here’s a list of some cool things you can do with this stuff:

• The table of California counties includes FIPS codes and polygons. FIPS codes can be used for linkage with US Census data, and polygons can be used to visualize county boundaries in Google Maps (or any other engine that supports KML).
• The table of counties and LEMSAs can be used to visualize California’s local EMS agencies with Google Maps.
• The table of California hospitals includes facilities which receive ambulance traffic in California. This table also includes each hospital’s identifiers for the California EMS Information System and various other systems held by the California Office of Statewide Health Planning and Development. (data linkage, anyone?!)
• The CEMSIS Hospital List 2011 includes hospitals along with SOME local EMS agency designations**, such as trauma centers, pediatric specialty centers, STEMI centers, and stroke receiving centers. **This list does not contain enough information to help you choose a hospital in case of emergency, so please don’t use it for that! Always call 911 in an emergency.
• If you survived that disclaimer, note the CEMSIS Hospital List is a great way to learn more about the key players in California’s Emergency Medical Services system.
• If you’re researcher working with OSHPD data, you can use the CEMSIS Hospital List 2011 to identify specialty hospitals as defined by local EMS agencies.

Next on the list are ambulance companies, fire departments, and more key players who take good care of us when we need them most.

Happy data visualizing =)

PS The spreadsheets and Google Fusion Tables are here.