In early January 2003, I received a call from a former colleague, Steve Kadish, who had just been appointed undersecretary of Health and Human Services for the incoming administration of Governor Mitt Romney in Massachusetts. Steve and I had worked together in Massachusetts government in the mid-90s, when he oversaw operational services for the state; my role was that of chief information officer for the state. Then from 1999 to 2002, we both worked on the turnaround of Harvard Pilgrim Health Care, an HMO. After that intense experience, I promised myself that I would do no more than one turnaround every 10 years or so!

But this promise was soon to be broken.

Steve was calling to ask if I would consider coming back to state government during a fiscal crisis to lead the IT operation of the state's Health and Human Services agency amid the most significant reorganization in its history. Doing so would put me back in the middle of a significant "turnaround", albeit in a public-sector context. And it would entail a jolting salary adjustment from the consulting work I was engaged in at the time. But it was an irresistible challenge to me, because of the tremendous opportunities in government to help individuals and communities, using information technology.

What followed is an adventure that resulted in something many people thought could not be accomplished - a virtual gateway to the 17 Health and Human Services (HHS) agencies in Massachusetts. For decades, these agencies operated separately and constructed separate systems within their own fiefdoms. Each had its own business processes (for common tasks like applying for benefits and eligibility determination) and used different forms to collect similar data. HHS was referred to as "the maze" by human services advocates.

Less than two months after the start of his administration in 2003, Governor Romney announced that as part of his plan for reorganizing HHS, he was going to build a virtual gateway to eliminate the maze that providers and beneficiaries had to navigate to work with HHS - at the time a $US12 billion - leviathan. In his proposal, the core administrative functions of HHS (IT, HR, finance) were to be consolidated, the HHS executive office would absorb the entire state Medicaid program, and the 17 agencies would be grouped into four core service areas.

We spent the first 10 months of 2003 thinking, planning and architecting an IT strategy for HHS. We focused on which business processes were most in need of improvement, and which improvements were most affordable. We singled out intake (filling out applications for benefits), eligibility determination and referral (letting applicants know of related services that may be helpful) as the processes we wanted to streamline and standardize first. Some states have attempted this kind of streamlining by replacing the underlying systems with newer, more comprehensive software. We lacked the money and felt too much time pressure to go in that direction. Instead, we decided to create a common portal to the 17 agencies, connecting the legacy systems with a service bus that would enable business processes to span systems.

Today, a single online electronic form provides access to the state's Medicaid, food stamps, WIC (the nutrition program for women, infants and children), subsidized child care, veterans services, care for the disabled and many other HHS services. Hospitals and community health centres in Massachusetts can now check for any pre-existing insurance coverage for low-income patients, and if none exists, immediately enter Medicaid application data online. This more inclusive coverage approach encourages preventive care rather than expensive emergency room care, and has reduced the burden to the public of paying for a large "uncompensated care" population.

This was a very complex undertaking, executed in a nearly impossible time frame. Even today, I wonder: What came together to make this happen?

Just-in-Time Development

As with any system design effort, particularly in government, scope creep can put extravagant demands on schedule and resources. After several months of somewhat fruitless attempts to get agreement on the exact dimensions of the gateway, we hit upon a strategy that worked very well: We decided to time the releases of the gateway using a rapid application design technique. Instead of asking "What is the ultimate scope?" and imagining an artificial completion date (with everyone's wish lists incorporated), we would identify the best set of services we could afford to build during the fiscal year.

We used a J2EE enterprise architecture to link the virtual gateway with the agencies' legacy systems. We chose to go in the direction of J2EE standards because they offered the greatest scalability and flexibility over the long term. We also chose to adopt a Web-based intake form application Deloitte Consulting had built for Pennsylvania. Deloitte performed the principal virtual gateway integration work, which entailed porting its application to a Java code base, and helping to orchestrate the connection of three existing systems - one using Cobol on the mainframe, another using Oracle on a mid-range platform, and the third a SQL server/.Net system - to an enterprise services bus implemented in BEA WebLogic.

The Gateway Goes Live

At 11.30 am on August 12, 2004, the virtual gateway release one (for intake, eligibility, and referrals for Medicaid, food stamps, WIC and child care) processed its first online application. Since then, the system has processed over 164,000 applications and averages 3000 applications per week. The first release cost less than $US8 million, including software, hardware and services. Subsequent releases have added intake for programs such as veterans and disabilities services and entirely different processes - for instance, helping with census management for homeless shelters.

In December 2005, an independent group known as Community Partners surveyed health-care providers about the virtual gateway. The survey found that the gateway not only made the job of front-line workers easier and significantly reduced the time necessary for an eligibility determination but also improved the experience of most Massachusetts residents in applying for benefits.

I can't say enough about the team that engaged in this effort. The development and operations teams worked virtually seven days a week for two months prior to the launch. And they're still going the extra mile. People poured energy into this project because they felt it mattered. For me, the project was especially instructive as an example of how an open, standard framework allows multiple systems to interoperate. Many opportunities of this sort are yet to be explored among government agencies. I have recently accepted the role of state CIO in hopes of doing so.

Louis Gutierrez, recently CIO of the Massachusetts Executive Office of Health and Human Services, is now CIO of the Commonwealth of Massachusetts, a position he also held from 1996 to 1998