If knowledge is power, ignorance is impotence. Citizens, consumers, investors, and patients all need trustworthy information when we vote, making purchasing decisions, buy stocks or other assets, or choose a surgeon, medical device, nursing home, or dialysis center. That’s why … Continue reading
The White House has launched COVIDTests.gov, which the Biden administration says will enable every home in the U.S. to order 4 free at-home COVID-19 tests through the mail, starting on January 19th — with no shipping costs or credit card required. Ideally, the administration will also allow Americans to request the high-quality masks President Biden said the US government would distribute through COVIDTests.gov as well.
As with Vaccines.gov, there’s a tremendous amount riding on the Biden administration delivering on this news service. Hundreds of millions of Americans REALLY need this administration to deliver on sending free tests and masks to the people through the mail who ask for them right now. This will be a simple but profound interaction.
If the White House can pull it off, it could rebuild public trust during a profoundly uneasy time, delivering the tests and masks that — with vaccines — would enable us all to navigate out of the pandemic together.
If COVIDTests.gov were to be overwhelmed by demand, flooding attacks, automated fraud, or test distribution is botched by the U.S. Postal Service, public trust could erode further.Last week, Politico reporter Ben Leonard reached out and asked a series of questions about whether the Biden administration would be able to deliver on a website to request rapid tests, raising concerns about whether this be a repeat of the Healthcare.gov debacle of almost a decade ago. The answers below lay out the case for why this White House is likely to succeed.
What were the key failures of healthcare.gov?
The public and major media focused a lot on the technology angle because healthcare.gov is a website, and it’s true that technical and design decisions caused major problems at the relaunch, when the Department of Health and Human Services moved it from being a glossy brochure to being a marketplace for health insurance. Lack of beta testing. Hosting at that couldn’t scale to meet demand. Incomplete integration between federal agency systems. Artificial bottlenecks in the marketplace flow.
That all led to a crisis when Americans began trying to use the site, in no small part because Healthcare.gov wasn’t iteratively built or tested “in the open” using modern software development practices, with the people it was meant to serve.
Much like the endemic IT failures that have bedeviled big state and federal projects for years, however, the fundamental problems stemmed from:
- Poor project management and oversight
- A government procurement system that builds and buys software like buildings and cars
- Outsourcing huge contracts to systems integrators and contractors
- Challenges recruiting and retaining technologists who must be sitting at the table from the beginning of a complex project
The team that rescued the site in the winter of 2013 was able to address many of these failures and then founded the U.S. Digital Service based upon these insights and the inspiration of Gov.uk in the United Kingdom.
- How can the Biden administration avoid them this time around?
The White House can give the U.S. Digital Service and 18F (the software development organization inside of the General Services Administration which could be involved) – resources and cover to do their best work. (Ideally, they will “show their work” as they go, too.) That means strong product management, iterative development,regular check-ins, designers and technologists in government, and working with best-in-class technology partners who understand how to build and scale modern responsive websites.
It’s also worth noting that building a website to request a COVID-19 test is not the same technical challenge as one that had to tie into the IRS to check eligibility for subsidies and complete a secure transaction. Failure would still be consequential, but the bar is much lower, as are the risks surrounding errors.
- What were the key issues with Trump’s promised national COVID-19 screening website? What lessons can be taken from it?
Former President Trump was unfit to lead a coordinated national response to a pandemic and uninterested in building out the national testing infrastructure that showed his lies about the prevalence of a deadly airborne virus to be false.
This promised website was vaporware, not a serious project that can or should be compared to past or present .gov efforts. The Trump White House never built or delivered anything after the President engaged in misleading hyperbole in the Rose Garden.
Instead of convening technologists, designers, and project managers from relevant agencies and private sector in a Manhattan Project for testing or grand national challenge, however, Trump misled the public by claiming that Google was already working on it. There’s no “there there” to compare.
- Vaccines.gov seemed to roll out more smoothly. Why do you think that happened and what lessons can be drawn?
I’d rack that up to:
- Competent leadership at the U.S. Digital Service involved from the outset in coordinating and managing the project, top-down cover from the Oval Office
- Subject matter expertise from the medical professionals who built VaccineFinder.org
- Deep technical expertise and capacity to deliver at scale from private sector companies involved
- All-out, mission-driven effort from many patriots inside and outside of government committed to connecting Americans with vaccines. A thread: https://twitter.com/digiphile/status/1388495731422543877
We all would know a lot more about what worked and why if the Biden administration had narrated its work in the open, held a press conference about Vaccines.gov, and taken questions instead of giving the news to Bloomberg on background.
- How big of a technical task do you think putting together this website will be?
If the functionality of COVIDTests.gov is limited to someone requesting a test be delivered to a given address, I don’t think that’s a big task for US government in 2022, even under heavy demand. If the COVIDTests.gov needs to authenticate someone and create accounts to prevent fraud or abuse, that will be a bit harder.
My hope is that we’ll see a lightweight website that uses Login.gov and a shortcode — say, text your zipcode to GETTST – that will enable people to quickly and easily request tests from a smartphone – along with a package of better, medical grade masks for themselves and their children that President Biden announced today would be made available for free to all Americans.
As an open enrollment deadline draws near, HealthCare.gov is once again having issues, according to a USA Today report.
The folks at Healthsherpa.com, however, tell me that it’s working fine. Last year, Ning Liang, one of the founders of the site, wrote me last year to tell me about updates to the site. Ling said that they can now enroll people in ACA marketplace plans, including subsidies.
According to Liang, as of June 2014 they were “the only place besides Healthcare.gov where this is possible. We have signed an agreement with CMS as a web based entity to do this. We are directly integrated with the federal data hub, so going through us is identical to going through Healthcare.gov.”
To date, they’ve helped more than 110,000 people sign up for insurance under the Affordable Care Act.
What a missed opportunity to have created a thriving ecosystem of other Web-based entities that are alternatives to the default government website.
Why did President Barack Obama go on “Between Two Ferns,” an irreverent talk show hosted by comedian Zack Galifiniakis?
The context comes about two thirds in, when the president talked about HealthCare.gov, young “invincibles” and health insurance that costs about as much as a cellphone bill: young people are still not aware of the policy or going to the site.
According to White House officials, the appearance on the show made FunnyOrDie.com, the comedy site that hosts the show, the number one referrer of Traffic to Healthcare.gov today.
— Tara McGuinness (@HealthCareTara) March 11, 2014
Update: According to an unnamed HHS official on a conference call today, some 19,000 of the 3 million people who have watched the video clicked on Healthcare.gov.
— John McCormack (@McCormackJohn) March 11, 2014
Update II: McGuiness tweeted that traffic to Healthcare.gov was up 40% compared to the previous day.
Funnyordie video has 11 million views. http://t.co/a7HUExG0vg traffic for yesterday was up almost 40% from Monday.
— Tara McGuinness (@HealthCareTara) March 12, 2014
In other words, it worked. I’ve asked for more comment from the White House on traffic and conversions and will update this post as necessary.
In the meantime, the answers to whether the president should find young audiences online, the White House should be making animated GIFs or the Department of Health and Human Services should be dabbling in doge memes are all a qualified yes: social media and online video are crucial mediums for outreach to a younger generation that has grown up digital, sharing what they read and watch with their friends and family.
Time Magazine’s’s sharp television critic, James Poniewozik, captured why the President of the United States going onto this particular show and adapting his delivery considerably from what Americans might see in other contexts worked better than, say, a Super Bowl interview:
It’s the tone of the comedy as much as the online medium that really targets the young audience Obama is pitching to here. There’s a cringe-humor generation gap; if you’re over a certain age, or simply haven’t watched much of a certain kind of contemporary comedy, you’ll probably watch it thinking that the segment is bombing and Obama is getting legitimately angry. But it’s a good fit for Obama’s sense of humor, which is a little dry and a little cutting–in ways that don’t always play in rooms when there are no ferns present.
The Obama administration announced significant adoption for the Blue Button in the private sector today. In a post at the White House Office of Science and Technology blog, Nick Sinai, U.S. deputy chief technology officer and Adam Dole, a Presidential Innovation Fellow at the U.S. Department of Health and Human Services, listed major pharmacies and retailers joining the Blue Button initiative, which enables people to download a personal health record in an open, machine-readable electronic format:
“These commitments from some of the Nation’s largest retail pharmacy chains and associations promise to provide a growing number of patients with easy and secure access to their own personal pharmacy prescription history and allow them to check their medication history for accuracy, access prescription lists from multiple doctors, and securely share this information with their healthcare providers,” they wrote.
“As companies move towards standard formats and the ability to securely transmit this information electronically, Americans will be able to use their pharmacy records with new innovative software applications and services that can improve medication adherence, reduce dosing errors, prevent adverse drug interactions, and save lives. ”
While I referred to the Blue Button obliquely at ReadWrite almost two years ago and in many other stories, I can’t help but wish that I’d finished my feature for Radar a year ago and written up a full analytical report. Extending access to a downloadable personal health record to millions of Americans has been an important, steadily shift that has largely gone unappreciated, despite reporting like Ina Fried’s regarding veterans getting downloadable health information. According to the Office of the National Coordinator for Health IT, “more than 5.4 million veterans have now downloaded their Blue Button data and more than 500 companies and organizations in the private-sector have pledged to support it.
As I’ve said before, data standards are the railway gauges of the 21st century. When they’re agreed upon and built out, remarkable things can happen. This is one of those public-private initiatives that has taken years to take fruit that stands to substantially improve the lives of so many people. This one started with something simple, when the administration gave military veterans the ability to download their own health records using from on MyMedicare.gov and MyHealthyVet and scaled progressively to Medicare recipients and then Aetna and other players from there.
There have been bumps and bruises along with the way, from issues with the standard to concerns about lost devices, but this news of adoption by places like CVS suggests the Blue Button is about to go mainstream in a big way. According to the White House, “more than 150 million Americans today are able to use Blue Button-enabled tools to access their own health information from a variety of sources including healthcare providers, health insurance companies, medical labs, and state health information networks.”
Notably, HHS has ruled that doctors and clinics that implement the new “BlueButton+” specification will be meeting the requirements of “View, Download, and Transmit (V/D/T)” in Meaningful Use Stage 2 for electronic health records under the HITECH Act, meaning they can apply for reimbursement. According to ONC, that MU program currently includes half of eligible physicians and more than 80 percent of hospitals in the United States. With that carrot, many more Americans should expect to see a Blue Button in the doctor’s office soon.
In the video below, U.S. chief technology officer Todd Park speaks with me about the Blue Button and the work of Dole and other presidential innovation fellows on the project.
One of the persistent concerns about Healthcare.gov regards the security of the federal health insurance exchange marketplace, as I reported for Politico Magazine this month. At least one glaring security flaw remained unpatched until the end of October. Despite the “big fix” announced on December 1, the security of the website and the backend systems behind it have not only remained in doubt, given issues that have come out in Congressional testimony but have now become the subject of contentious exchanges between the United States House Oversight Committee and the Department of Health and Human Services, which operates them.
Today, Democrats on the House Energy and Commerce Committee released a memorandum regarding a security briefing on the Affordable Care Act (embedded below) that includes a summary of a classified briefing from Dr. Kevin Charest, the HHS Chief Information Security Officer, and Ned Holland, HHS Assistant Secretary for Administration. The memorandum states that “there have been no successful security attacks on Healthcare.gov. In it, Dr. Charest is quoted as saying that “no person or group has hacked into Healthcare.gov, and no person or group has maliciously accessed any personally identifiable information from users.”
The authors of the memorandum, Representatives Henry A. Waxman and Diana DeGette, write that “the information provided in the briefing was reassuring,” given the assurances of the chief information security officer that “the security of Healthcare.gov has not been breached, and hackers have had no access to personally identifiable information.”
Despite this letter, it’s not clear whether the Healthcare.gov security concerns that TrustedSec has highlighted have been addressed. Given the continued focus of Congressional committees on the issue, expect more assessments and audits to emerge in the future.
UPDATE: In June 2014, Ning Liang, one of the founders of HealthSherpa, wrote in about updates to the site. Ling said that they can now enroll people in ACA marketplace plans, including subsidies. According to Liang, “we are the only place besides Healthcare.gov where this is possible. We have signed an agreement with CMS as a web based entity to do this. We are directly integrated with the federal data hub, so going through us is identical to going through Healthcare.gov.”
Earlier tonight, Levick director of digital content Simon Owens discovered HealthSherpa.com, thought it was cool, and read a Daily Dot post about it that framed it as 3 20-something San Francisco Bay-area resident coding up an alternative to Healthcare.gov.
Could it be that easy, wondered Owens? Could these young coders have created a simpler, better way to shop for health insurance than the troubled Healthcare.gov?
Well, yes and no. As is so often the case, it’s not quite that simple, for several reasons.
1) As always, note the disclaimer at the bottom of HealthSherpa.com: “The information provided here is for research purposes. Make sure to verify premiums and subsidies on your state exchange or healthcare.gov, or directly with the insurance company or an agent.”
Why? The site is based upon the publicly available data published by the Department for Health and Human Services, individual state exchanges and Healthcare.gov for premium costs, like this dataset of premiums by county at data.healthcare.gov.
— The Health Sherpa (@healthsherpas) November 10, 2013
Unfortunately, there appear to be data quality issues, as CBS News reported, that may be an issue on both sites.
— The Health Sherpa (@healthsherpas) November 4, 2013
When I compared searches for the same zipcode in Florida for a 35 year old, single non-smoking male, I found the same 106 plans but was quoted different premiums: $128.85 on HC.gov vs $150.24 on HealthSherpa. Hmm.
That could be user error, but… it looks like Healthcare.gov continues to underestimate costs.
Healthsherpa may actually be doing better, here. Good job, guys.
2) Regardless, this is not a replacement for everything Healthcare.gov is supposed to do.
The federal and state exchanges aren’t just about browsing plans and comparing premiums for options in a given zipcode in the “marketplace.” After a user knows decides which plan he or she want, the software is supposed to:
1) Register them as a user (registration was up front initially, which was a controversial, important choice, relevant to the site crashing at launch)
2) Authenticate them against government data bases
3) Verify income against government data bases
4) Calculate relevant subsidies, based upon income
5) Guide them through the application process
6) Send that form data on to insurance companies for enrollment.
The tech that underpins the test and graphics website on the front end of those process continues to hold up well.
The rest of the software that is supposed to enable visitors to go through steps 1-6 software, not so much. 16 state exchanges and DC are having varying degrees of success, with HHS Secretary Kathleen Sebelius acknowledging issues with data quality in Step 6 in her testimony to Congress.
3) While it has a subsidy calculator, otherwise Healthsherpa doesn’t replace Healthcare.gov.
Healthsherpa enables you to find a relevant plan and then gives you contact info for the relevant insurer.
“Call Humana Medical Plan, Inc. at (800) 448-6262.
Use their menu or ask the operator to speak to someone about purchasing coverage.
Tell them you would like to purchase health exchange coverage, specifically the Humana Connect Basic 6350/6350 Plan for Hillsborough County, FL.
Follow their instructions to complete the application process.”
It does not place calls to the data hub to calculate steps 1-6.
That limited functionality, however, has been good enough for U.S. Senator Angus King to recommend HealthSherpa as a temporary alternative to Healthcare.gov.
“HealthSherpa offers a user-friendly platform to quickly browse through available health insurance plan options, including monthly premium costs, coverage plans, and possible premium subsidies,” Senator King said. “I recommend that Mainers who are having trouble with Healthcare.gov use HealthSherpa as a temporary alternative until the federal website functions properly.”
4) There are OTHER private healthcare insurance brokers that could be doing this.
Back in May 2013, the Center for Medicare and Medicaid Services issued official guidance for private sector brokers in online health insurance marketplaces. (PDF)
Former U.S. chief technology officer Aneesh Chopra said that these “Web-based entities” will be online this fall, operated by entities like eHealthInsurance.com and GetInsured.
For some reason, however, private sector insurance brokers have been stymied by the federal government from selling ACA insurance policies.
That’s unfortunate, given that the Obama administration could use a Plan B, just in case the progress on Healthcare.gov doesn’t lead to a functional federal health insurance exchange twenty days from now.
Update: Jonathan Cohn, writing for the New Republic, looked into Healthcare.gov’s backup plan and comes up with an interesting detail: issues with the so-called data hub could be holding back deployment of private online health insurance brokers.
…administration officials have been huddling with insurers about how to make more use of direct enrollment. Step one is to make sure that “side door” enrollment works smoothly. It doesn’t function well right now, because—you guessed it—it relies on the same information technology system that powers healthcare.gov. Fixing that portal, which techies tell me is called an “application programming interface,” is high on the administration’s to-do list. But it’s not clear (to me) whether improving the portal might require design modifications—or to what extent its success depends upon other, ongoing repairs to the federal website.
So, here’s some speculation: While it’s hard to know for sure, but it’s quite likely that that “portal” is the data hub that’s behind Healthcare.gov, and that it may not be up to additional volume from private sector demand.
The federal exchange and state exchanges both rely upon it, and, while federal officials have said that it’s working, a report by the New York Times yesterday that some state health insurance exchange are continuing to battle tech problems indicated that it’s not holding up under demand:
Even states whose websites are working well say they are hampered by a common problem: the federal website, particularly the data hub that checks every applicant’s identity and eligibility. That hub has stopped working on several occasions, preventing applications in the states from being completed.
If that’s happening now, concerns about the ability of the hub to hold up under the pressure of private sector online insurance brokers could well be justified. If I learn anything more definitive, I’ll share it.
Over the past 27 days, as I’ve steadily shared analysis and links on what went wrong in the botched re(launch) of Healthcare.gov on Twitter, Facebook and Google+, I’ve also been talking in more depth about what went wrong on various media outlets, including:
- The Kojo Nnamdi Show on WAMU: “Healthcare.gov: what went wrong?” [LISTEN] [TRANSCRIPT]
- Radio Times on WHYY: “What’s up with the federal health insurance exchange?” [LISTEN]
- Washington PostTV: “The bumpy launch of healthcare.gov“
- Fox News Radio: Technical issues at Healtcare.gov (archived here)
Last week, the Obama administration announced a plan to fix the issues with the software behind HealthCare.gov, including putting QSSI in charge of as the “general contractor” and prioritizing fixing errors in 834 file data first, with the goal of have the system functioning end-to-end by November 30.
The teams of Presidential Innovation Fellows and “A List” contractors in the “tech surge” to fix the software have a tough challenge ahead of them. According to reporting from The New York Times and The Washington Post, Healthcare.gov wasn’t tested as a complete system until the last week of September, when it crashed with only a few hundred users.
Despite the issues revealed by this limited testing, government officials signed off on it launching anyway, and thus was born a historic government IT debacle whose epic proportions may still expand further.
Should the White House have delayed?
“When faced with go live pressures, I tell my staff the following:
‘If you go live months late when you’re ready, no one will ever remember. If you go live on time, when you’re not ready, no one will ever forget.”-Dr. John Halamka, CIO Beth Israel Deaconness Hospital
In retrospect, the administration might have been better served by not launching on October 1st, something that was within HHS Secretary Kathleen Sebelius’ legal purview. After all, would the federal government launch a battleship that had a broken engine, faulty wiring or non-functional weapons systems into an ongoing fight? This software wasn’t simply “buggy” at launch — It was broken. These weren’t “glitches” caused by traffic, although the surge of traffic did expose where the system didn’t work quickly. Now that a reported 90% of users are able to register, other issues on the backend that are just beginning to become clear, from subsidy calculation to enrollment data to insurers reporting issues with what they are receiving to serious concerns about system security.
Based upon what we know about troubles at Healthcare.gov, it appears that people from the industry that were brought in to test the Healthcare.gov system a month ago urged CMS not to go live. It also appears that inside the agency who saw what was going on warned leadership months in advance that the system hadn’t been tested end-to-end. Anyone building enterprise software should have the code locked down in the final month and stopped introducing new features 3-6 months prior. Instead, it appears new requirements kept coming in and development continued to the end. The result is online now. (Or offline, as the case may be.)
On September 30, President Obama could have gone before the American people and said that the software was clearly not ready, explained why and told Americans that his administration wouldn’t push it live until they knew the system would work. HHS could have published a downloadable PDF of an application that could be mailed in and a phone number on the front page, added more capacity to the call centers and paper processing. It’s notable that three weeks later, that’s pretty much what President Obama said they have done.
The failed launch isn’t just about “optics,” politics or the policy of the Affordable Care Act itself, which is a far greater shift in how people in the United States browse, buy, compare and consume health insurance and services. A working system represents the faith and trust of the American people in the ability of government. This is something Jennifer Pahlka has said that resonates: how government builds websites and software matters, given the expectations that people now have for technology. The administration has handed the opponents of the law an enormous club to bash them with now — and they’ll deserve every bit of hard criticism they get, given this failure to execute on a signature governance initiative.
Articles worth reading on Healthcare.gov and potential reforms
- Whenever I see Fred Trotter, I’m reminded that he’s forgotten more about open source software and healthcare IT than I’m ever likely to learn. Last week, he talked with Ezra Klein about the issues with Healthcare.gov
- Ezra Klein also talked to Clay Johnson about the lessons of Healthcare.gov (hint: procurement, project management and insourcing)
- Tough reporting on failures in e-government is critical to improving those services for all, but particularly for the poor.
- A post by Development Seed founder Eric Gunderson on the open source front-end for Healthcare.gov: “It’s called Jekyll, and it works.”
- Rusty Foster on Healthcare.gov: it could have been worse. This failure to (re)launch just happened under vastly more political scrutiny and deadlines set by Congress. The FBI’s Sentinel program, by contrast, had massive issues — but you didn’t see the Speaker of the House tweeting out bug reports or cable news pundits opining about issues. The same is true of many other huge software projects.
- A must-read op-ed by former Obama campaign CTO Harper Reed and Blue State Digital co-founder and former Presidential Innovation Fellow Clay Johnson on what ails government IT, adding much-needed context to what ailed Healthcare.gov
- A Mother Jones interview that asked whether Reed and other former campaign staff could fix Healthcare.gov. (Spoiler: No.)
- If not those folks, then how should the administration fix Healthcare.gov? In the larger sense, either the federal government will reform how it buys, builds and maintains software, through a combination of reforming procurement with modular contracting, bringing more technologists into government, and adopting open source and agile development processes …or this will just keep happening. The problems go much deeper that a “website.”
- Ezra Klein pulled all of these pieces together in a feature on the “broken promise of better government through technology” at the end of the month. (He may have been heard in the Oval Office, given that the president has said he reads him.) Speaking at an “Organizing for America” event on November 4th, President Obama acknowledged the problem. “…I, personally, have been frustrated with the problems around the website on health care,” he said, “And it’s inexcusable, and there are a whole range of things that we’re going to need to do once we get this fixed – to talk about federal procurement when it comes to IT and how that’s organized…”
- The issues behind Healthcare.gov cannot only be ascribed to procurement or human resources, as Amy Goldstein and Juliet Eilperin reported in the Washington Post: insularity and political sensitivity were a central factor behind the launch..
Based on interviews with more than two dozen current and former administration officials and outsiders who worked alongside them, the project was hampered by the White House’s political sensitivity to Republican hatred of the law — sensitivity so intense that the president’s aides ordered that some work be slowed down or remain secret for fear of feeding the opposition. Inside the Department of Health and Human Services’ Centers for Medicare and Medicaid, the main agency responsible for the exchanges, there was no single administrator whose full-time job was to manage the project. Republicans also made clear they would block funding, while some outside IT companies that were hired to build the Web site, HealthCare.gov, performed poorly.
- What could be done next? Congress might look across the Atlantic Ocean for an example. After one massive IT failure too many, at the National Health Service the United Kingdom created and empowered a Government Digital Services team. UK Executive Director of Digital Mike Bracken urged U.S. to adopt a digital core.”
- In the video below, Clay Johnson goes deep on what went wrong with Healthcare.gov and suggests ways to fix it.
- Can the White House and Congress take on the powerful entrenched providers in Washington & do the same? I’m not optimistic, unfortunately, given the campaign contributions and lobbying prowess of those entities, but it’s not an impossible prospect. I’ll write more about it in the future.
UPDATE: The refresh of Healthcare.gov in June went well. On October 1st, when the marketplace for health insurance went live at the site.gov, millions of users flocked to the website and clicked “apply now.” For days, however, virtually none of them were able to create accounts, much less complete the rest of the process and enroll for insurance. By the end of the week, however, it was clear that the problems at Healthcare.gov were not just a function of high traffic but the result of the failure of software written by private contractors, with deeper issues that may extend beyond account creation into other areas of the site. On October 9th, as prospective enrollees continued to be frustrated by error-plagued websites around the country, I joined Washington Post TV to give a preliminary post-mortem on why the HealthCare.gov relaunch went so poorly.
The article that follows, which was extended and published at The Atlantic, describes the team and process that collaborated on launch of the new site in June, not the officials or contractors that created the botched enterprise software application that went live on October 1st. In the Atlantic, I cautioned that “…the site is just one component of the insurance exchanges. Others may not be ready by the October deadline.” The part of the site I lauded continues to work well, although the Github repository for it was taken offline. The rest has …not. I’ve taken some heat in the articles’ comments and elsewhere online for being so positive, in light of recent events, but the reporting holds up: using Jekyll is working. Both versions of the story, however, should have included a clearer caveat that the software behind the website had yet to go live — and that reports that the government was behind on testing Healthcare.gov security suggested other issues might be present at launch. If readers were misled by either article, I apologize. –Alex
Healthcare.gov already occupies an unusual place in history, as the first website to be demonstrated by a sitting President of the United States. In October, it will take on an even more important historic role, guiding millions of Americans through the process of choosing health insurance.
How a website is built or designed may seem mundane to many people in 2013, but when the site in question is focused upon such a function, it matters. Yesterday, the United States Department of Health and Human Services (HHS) relaunched Healthcare.gov with a new look, feel and cutting edge underlying architecture that is beyond rare in federal government. The new site has been built in public for months, iteratively created by a team of designers and engineers using cutting edge open source technologies. This site is the rarest of birds: a next-generation website that happens to be a .gov.
“It’s fast, built in static HTML, completely scalable and secure,” said Bryan Sivak, chief technology officer of HHS, in an interview. “It’s basically setting up a Web server. That’s the beauty of it.”
The people building the new Healthcare.gov are unusual: instead of an obscure sub-contractor in a nameless office park in northern Virginia, a by a multidisciplinary team at HHS worked with Development Seed, a scrappy startup in a garage in the District of Columbia that made its mark in the DC tech scene deploying Drupal, an open source content management system that has become popular in the federal government over the past several years.
“This is our ultimate dogfooding experience,” said Eric Gundersen, the co-founder of Development Seed. “We’re going to build it and then buy insurance through it.”
“The work that they’re doing is amazing,” said Sivak, “like how they organize their sprints and code. It’s incredible what can happen when you give a team of talented developers and managers room to work and let them go.”
What makes this ambitious experiment in social coding unusual is that the larger political and health care policy context that they’re working within is more fraught with tension and scrutiny than any other arena in the federal government. The implementation and outcomes of the Patient Protection and Affordable Care Act — AKA “Obamacare” — will affect millions of people, from the premiums they pay to the incentives for the health care they receive.
“The goal is get people enrolled,” said Sivak. “A step to that goal is to build a health insurance marketplace. It is so much better to build it in a way that’s open, transparent and enables updates. This is better than a big block of proprietary code locked up in CMS.”
The new Healthcare.gov will fill a yawning gap in the technology infrastructure deployed to support the mammoth law, providing a federal choice engine for the more than thirty different states that did not develop their own health insurance exchanges. The new website, however modern, is just one component of the healthcare insurance exchanges. Others may not be ready by the October deadline. According to a recent report from the Government Accountability Office, the Department of Health and Human Services’ Centers for Medicare & Medicaid Services (CMS) is behind in implementing key aspects of the law, from training workers to help people navigate the process to certifying plans that will sold on the exchanges to determining the eligibility of consumers for federal subsidies. HHS has expressed confidence to the GAO that exchanges will be open and functioning in every state on October 1.
On that day, Healthcare.gov will be the primary interface for Americans to learn about and shop for health insurance, as Dave Cole, a developer at Development Seed, wrote in a blog post this March. Cole, who served as a senior advisor to the United States chief information officer and deputy director of new media at the White House, was a key part of the team that moved WhiteHouse.gov to Drupal. As he explained, the code will be open in two important ways:
First, Bryan pledged, “everything we do will be published on GitHub,” meaning the entire code-base will be available for reuse. This is incredibly valuable because some states will set up their own state-based health insurance marketplaces. They can easily check out and build upon the work being done at the federal level. GitHub is the new standard for sharing and collaborating on all sorts of projects, from city geographic data and laws to home renovation projects and even wedding planning, as well as traditional software projects.
Moreover, all content will be available through a JSON API, for even simpler reusability. Other government or private sector websites will be able to use the API to embed content from healthcare.gov. As official content gets updated on healthcare.gov, the updates will reflect through the API on all other websites. The White House has taken the lead in defining clear best practices for web APIs.
Thinking differently about a .gov
According to Sivak, his team didn’t get directly involved in the new Healthcare.gov until November 2012. After that “we facilitated the right conversations around what to build and how to build it, emphasizing the consumer-facing aspects of it,” he said. “The other part was to figure out what the right infrastructure was going to be to build this thing.”
That decision is where this story gets interesting, if you’re interested in how government uses technology to deliver information to the people it serves. Government websites have not, historically, been sterling examples of design or usability. Unfortunately, in many cases, they’ve also been built at great expense, given the dependence of government agencies on contractors and systems integrators, and use technologies that are years behind the rest of the Web. Healthcare.gov could have gone in the same direction, but for the influence of its young chief technology officer, an “entrepreneur-in-residence” who had successfully navigated the bureaucracies of the District of Columbia and state of Maryland.
“Our first plan was to leverage Percussion, a commercial CMS that we’d been using for a long time,” said Sivak. “The problem I had with that plan was that it wasn’t going to be easy to update the code. The process was complicated. Simple changes to navigation were going to take a month.”
At that point, Sivak did what most people do in this new millennium when making a technology choice: he reached out to his social networks and went online.
“We started talking to people about a better way, including people who had just come off the Obama campaign,” he said. “I learned about the ground they had broken in the political space, from A/B testing to lightweight infrastructure, and started reading about where all that came from. We started thinking about Jekyll as a platform and using Prose.io.”
After Sivak and his team read about Development Seed’s work with Jekyll online, they contacted the startup directly. After a little convincing, Development Seed agreed to do one more big .gov project.
“A Presidential Innovation Fellow used same tech we’re using for several of their projects,” said Cole. “Bryan heard about it and talked to us. He asked where we would go. We wanted to be on Github. We knew there were performance and reliability benefits from building the stack on HTML.”
Jekyll, for those who are unfamiliar with Web development trends, is a way for developers to build a static website from dynamic components. Instead of running a traditional website with a relational database and server-side code, using Jekyll enables programmers to create content like they create code. The end result of this approach is a site that loads faster for users, a crucial performance issue, particularly on mobile devices.
That adds up to cost savings. Sites that are heavily trafficked — as Healthcare.gov can reasonably be expected to be – normally have to use a caching layer to serve static content and add more server capacity as demand increases.
“When we worked with the World Bank, they chose a plan from Rackspace for 16 servers,” said Gundersen. “That added tens of thousands of dollars, with a huge hosting bill every month.”
HHS had similar strategic plans for the new site, at least at first.
“They were planning 32 servers, between staging, production and disaster recovery, with application servers for different environments,” said Cole. “You’re just talking about content. There just needs to be one server. We’re going to have 2, with one for backup. That’s a deduction of 30 servers.”
While Jekyll eliminates the need for a full-blown content management system on the backend of Healthcare.gov (and with it, related costs), the people managing the site still need to be able to update it. That’s where Prose.io comes in. Prose.io is an open source content editor created by Development Seed that gives non-programmers a clean user interface to update pages.
“If you create content and run Jekyll, it requires content editors to know code,” said Cole. “Prose is the next piece. You can run it on your on own servers or use a hosted version. It gives access to content in a CMS-like interface, basically adding a WYSIWYG skin, giving you a text editor in the browser.”
In addition to that standard “what you see is what you get” interface, familiar from WordPress or Microsoft Word, Prose.io offers a couple of bells and whistles, like mobile editing.
“You can basically preview live,” said Cole. “You usually don’t get a full in-browser preview. The difference is that you have that with no backend CMS. It’s just a directory and text files, with a Web interface that exposes it. There are no servers, no infrastructure, and no monthly costs. All you need is a free Web app and Github. If you don’t want to use that, use Git and Github Enterprise.” Update: Cole wrote more about launching Healthcare.gov on the DevelopmentSeed blog on Tuesday.
Putting open source to work
Performance and content management aside, there’s a deeper importance to how Healthcare.gov is being built that will remain relevant for years to come, perhaps even setting a new standard for federal government as a whole: updates to the code repository on Github can be adopted for every health insurance exchange using the infrastructure. (The only difference between different state sites is a skin with the state logo.)
“We have been working in the .gov space for a while,” said Gundersen. “Government people want to make the right decisions. What’s nice about what Bryan is doing is that he’s trying to make sure that everyone can learn from what HHS is doing, in real-time. From a process standpoint, what Bryan is doing is going to change how tech is built. FCC is watching the repository on Github. When agencies can collaborate around code, what will happen? The amount of money we have the opportunity to save agencies is huge.”
Collaboration and cascading updates aren’t an extra, in this context: they’re mission-critical. Sivak said that he expects the new site to be improved iteratively over time, in response to how people are actually using it. He’s a fan of the agile development methodology that has become core to startup development everywhere, including using analytics tools to track usage and design accordingly.
“We’re going to be collecting all kinds of data,” said Sivak. “We will be using tools like Optimizely to do A/B and multivariate testing, seeing what works on the fly and adapting from there. We’re trying to treat this like a consumer website. The goal of this is to get people enrolled in health care coverage and get insurance. It’s not simple. It’s a relatively complex process. We need to provide a lot of information to help people make decisions. The more this site can act in a consumer-friendly fashion, surfacing information, helping people in simple ways, tracking how people are using it and where they’re getting stuck, the more we can improve.”
Using Jekyll and Prose.io to build the new Healthcare.gov is only the latest chapter in government IT’s quiet open source evolution. Across the federal government, judicious adoption of open source is slowly but surely leading to leaner, more interoperable systems.
“The thing that Git is all about is social coding,” said Sivak, “leveraging the community to help build projects in a better way. It’s the embodiment of the open source movement, in many ways: it allows for truly democratic coding, sharing, modifications and updates in a nice interface that a lot of people use.”
Open by design
Sivak has high aspirations, hoping that publishing the code for Healthcare.gov will lead to a different kind of citizen engagement.
“I have this idea that when we release this code, there may be people out there who will help us to make improvements, maybe fork the repository, and suggest changes we can choose to add,” he said. “Instead of just internal consultants who help build this, we will suddenly have legions of developers.”
Not everything is innovative in the new Healthcare.gov, as Nick Judd reported at TechPresident in March: the procurement process behind the new site is complicated and the policy and administrative processes that undergird it aren’t finished yet, by any account.
The end result, however, is a small startup in a garage rebuilding one of the most important federal websites of the 21st century in a decidedly 21st century way: cheaper, faster and scalable, using open source tools and open standards.
“Open by design, open by default,” said Sivak. “That’s what we’re doing. It just makes a lot of sense. If you think about what should happen after this year, all of the states that didn’t implement their systems, would it make sense for them to have code to use as their own? Or add to it? Think about the amount of money and effort that would save.”
That’s a huge win for the American people. While the vast majority of visitors to Healthcare.gov this fall will never know or perhaps care about how the site was built or served, the delivery of better service at lowered cost to taxpayers is an outcome that matters to all.
This Monday, Minister of Parliament Jeremy Hunt, the United Kingdom’s Secretary of State for Health, delivered a keynote address at the fourth annual Health Datapolooza in Washington, DC. In a rhetorical turn that would be anathema for any national conservative politician on this side of the Atlantic, Hunt commended the United States for taking steps towards providing universal health insurance to its people.
— Alex Howard (@digiphile) June 3, 2013
Hunt outlined three major elements in a strategy to improve health care in the UK: 1) applying data more effectively 2) improving transactional capabilities and 3) putting patients in the “driver’s seat” of their own health care. He pointed to several initiatives that support that strategy, from extending electronic health records to 50 million people to sequencing the genomes of 100,000 people and developing telemedicine capabilities for 3 million patients. Given the focus of the datapalooza, however, perhaps his most interesting statement came with respect to personal data ownership:
— Alex Howard (@digiphile) June 3, 2013
After the keynote, I interviewed Secretary Hunt and Tim Kelsey, the first national director for patients and information in the National Health Service. Our discussion, lightly edited, follows.
What substantive steps has the UK taken to actually putting health data in the hands of patients?
Hunt: Basically, I have given an instruction that everyone should be able to access their own health record online before the next general election, which means that I will be accountable for delivering that promise. There’s no wiggle room for me. That’s a big change, and it’s also a big change for the system because, basically, it means that every hospital and every general practitioner has to get used to the idea that the data they write about patients will be able to be accessed by patients. It’s a small but very significant first step.
There’s sometimes a disconnect between what politicians direct and what systems actually do. What’s happening with the UK’s long-delayed EHR system?
Hunt: I’ve given a pretty accountable timeframe for this: May 2015. I’ll be facing a general election campaign then. If we don’t deliver, then my head’s going to be on the block. I think it is a valid question, because of course once you set these objectives, then you start to look underneath it. One of the questions that we have to ask ourselves is how many have actually used this. We want everyone to be able to use this, but in practice, if the way they use it is they’re going to have to go into their GP, they’ve got to sign a consent form, there’s some complex procedure, then actually it’s not going to change people’s lives. The next question is about take-up, and that’s what we’re exploring at the moment.
Are there any aspects of the U.S. healthcare system that you think might be worth adopting and bringing back to the U.K.? Or vice versa?
Well, it’s quite interesting. We just had a really good meeting with [US CTO] Todd Park. I don’t think the differences are so great. I mean, on one level, yes, hospitals here are private or charitable, so they can’t be mandated by the government to do anything. And yet, they’ve succeeded in getting 80% of them to adopt EHRs through setting a standard and a certain amount of financial incentive. We can tell our hospitals to do things, but actually, as you said earlier, that’s not the same as them actually doing it.
I think in the end, in both countries, what you have to do is make it so that it’s in the hospitals’ own interest. In our case, the way that we’re doing that is trying to demonstrate that sensibly embracing the technology agenda has a massive effect on reducing mortality rates and improving clinical outcomes. By publishing all of the data about those outcomes, we’re creating competition between hospitals. That, I hope, will drive this agenda.
At the same time, we need to change public awareness. This is the big challenge – this sense that you can actually be in charge of your own health is just, surprisingly, absent in large numbers of people. There’s a very strong sense that lots of people have that “health is something that’s done to me” by NHS.
In the U.S has released data on the disparities in pricing for hospital procedures and comparisons of hospital quality — but you still need to go to places that take your health insurance. In the NHS, is that as much of an issue?
Hunt: That’s a really good question to ask because, in the U.K, for virtually any procedure, you have the right to have it done in any hospital in the country — and yet, very few people avail themselves of that right. So, by publishing surgical survival rates, we’re hoping to create pressure, where people actually say “I’m going to have this heart operation, and I’m not going to go to my local hospital, I’m going to go to this one a bit farther away that has higher success rates.” At the moment, people don’t actually do that; they tend to go where they’re recommended to. That’s where this information revolution can take hold.
What is the most unexpected thing that has happened since the U.K. began releasing more open data about health?
Tim Kelsey: I don’t know if this is unexpected or not, but the most startling thing is that we’ve moved from having one of the worst heart surgery survival rates in Europe to being the best. Heart surgery is the only speciality where we’ve published comparative data by heart surgeons across the whole country.
Do you think that’s an accident?
Tim Kelsey: No, I don’t think it’s an accident at all. Within that data, if you look at what has actually happened, the assumption of the geniuses who actually pioneered the program was that the gap between the best surgeons and the worst surgeons would narrow, because the weaker surgeons would raise their game. That didn’t happen. What happened was that the best surgeons got even better, and the underperforming surgeons also raised their game. The truth is that they want to be the winner, and open data has had a massive impact in driving outcomes and standards.
What are the most important principles or substantive steps that you’re applying at the NHS to mitigate risks or harms from privacy breaches?
Hunt: We have to carry the public with us. We have a very strong free press, as you do, and we’re very proud of that. If they believe that people’s data is going to be used to infringe their privacy, then public confidence in the huge revolution that the dataaplooza is all about will be shaken and lack a massive impact. I think that there’s a very simple way that you maintain public confidence, which is by making it absolutely clear that you own that data. You can choose, if you don’t want that data to be used, in even in an anonymized form, you can say I’m not going to share my data. I think once you do that, you create a discipline in the system to make sure that the anonymization of data is credible, because people can withdraw their consent if they don’t believe it.
Also, you put people in the driver’s seat, because I think people’s motives are different. You and I, as young and hopefully healthful individuals, we’re thinking about privacy. If somebody’s got terrible cancer, he’s actually thinking, ‘well, I would really like my data to be used for the benefit of humanity.’ They’re actually very, very happy to have their data shared. They have a different set of concerns.
I don’t think you’ll have any trouble, for example, getting 100,000 people to consent to have their genome sequenced. These will be people who have cancer, and once you have cancer, you think, ‘what can I do to help future generations conquer cancer?’ The mentality changes. We have to maintain people’s confidence.
I think the best analogy, though, is banking. Perhaps the second thing people care about most after their health is their money, and the banks have been able to maintain people’s confidence. They’re actually doing banking online, so that you can access your bank account from any PC, anywhere in the world. It’s something you can do with confidence. They’ve done that because they’ve thought through the procedures.
In the U.S., you’re entitled to access a free copy of your credit report once a year. Consumers, however, still don’t have access to their own data across much of the private sector. Will the British government support “rights to data for its citizens?”
Hunt:: We are hoping to preempt the worry about that by instructing the NHS that everyone has a right of veto over the use of their own data. You own your own medical record. If you don’t want that shared, then that’s your decision, and you’re able to do that. If we didn’t do that, I think the courts might make us do that.
Kelsey: Just to clarify that point: The Data Protection Act, which is effectively a European piece of legislation, says that people have the right to object to data being shared, in any context, private sector or health or otherwise, or to opt out. We’ve said, because of the rights priority we’re giving to patients as the de facto owner of the data, which is different from the American situation so far.
We’re setting a global standard here, which will be interesting experiment for the rest of the world to watch, that people will have the right to say “I don’t want my data shared” — and people will respect that. Now, at the moment that is not a legal right, that is a de facto right that will be expected. It may well be that we’ll need to simply write down a law that this is an individual’s data and rights flow from that. At the moment, there’s no law that gives an individual patient the right to their own data nor to opt out out of its sharing.