About Health Satisfaction Two Point Oh!

The past year has seen some leaps and bounds in the way me and my team can rapidly handle publicly reportable data and get it onto Web sites. The two underpinnings of this are CLAIRE, the Claire Lightweight Agile IPRO Reporting Environment; and Pellucid, the health care transparency database system. CLAIRE is a software framework built on CakePHP that allows the dev team to grab up Pellucid data and quickly deploy Web applications. Pellucid is a MySQL data warehouse we built to house every single publicly reportable value we know of or can calculate in house. More on both of these technologies later.

So last week I initiated a test. I gave the dev team a week to come up with a complete rewrite of our Hospital Satisfaction Web site, and lo and behold they did it! You can go visit AboutHealthSatisfaction.org and review patient surveys of nearly every hospital in the US! Why are you still here? Click the link already.‎ Full story...

Why Not The Second Best?

'Tis the season to work all night every night for a couple of weeks. The newly updated WhyNotTheBest.org went live this weekend, with a spiffy new look and four gajillion new rows of data.

Well, truth be told it's about six million rows. But it's still a lot. WhyNotTheBest is a Web property funded by the Commonwealth Fund that is geared toward health care providers and policy makers, the basic intent being to identify the top performers for a range of measures and allow anyone to compare (benchmark) themselves against these top performers.

The new design is thanks to Digital Wave who laboured over a very functional search engine that allows any user to add as many hospitals to a comparison list as they want.

The new site contains a bunch of new data, notably the readmission and mortality rates published by CMS for several conditions, and we included average Medicare reimbursements for the same conditions.

In addition we added a whole slew of new filters, so it is now possible to create groups of hospital by Dartmouth HRR (Hospital Referral Region), by ownership, by health system and a whole lot more.

As always the data drives the user to improvement tools and intervention material that will help improve quality of care for the measures being published.

Please go check it out and comment here if you have some feedback!‎ Full story...

No Ma'am, We're Statisticians

Yes, it's true. I wanted to build the Illinois Health Care Report Card just so I could finally write a blog post laden with Blues Brother quotes. You could say I was on a mission from God. Our Lady of Blessed Web Site Acceleration don't fail us now.

Jokes aside, the IPRO eServices team has never worked this hard. Incidentally, the work allowed me to visit Chicago a few times and I have to say it is a surprisingly fun place to be, even if you can't find the Cook County Assessor's Office. We had just a few short months to get Illinois up and running, and by today's numbers I think we did a great job.

We launched early this morning and traffic started flooding in like I've never seen. Let's have a look at what's in it.

Quality

Quality data leads the tabs, and we populated it with both process and outcomes measures. We took the process of care measures from hospitalcompare.gov and rolled them up into three neat aggregate scores for Heart Attack, Heart Failure and Pneumonia. Following that are six mortality measures using AHRQ Inpatient Quality Indicators. Unfortunately due to some complex issues surrounding the measures, we couldn't risk adjust the data this quarter so the data are observed rates for now, we'll fix that for the New Year. We round off the tab with some utilisation data including the C-section rate.

Safety

For safety we published a range of Surgical Care Improvement measures, then we get the much awaited Hospital-Acquired Infection data. Illinois is out in front with collecting this data, and we had some good stories to tell, but we are looking forward to hospital-level MRSA rates next year. For now you can review central line-associated bloodstream infection rates. The last module on the safety tab shows a selection of the Patient Safety Indicators from AHRQ including accidental puncture/laceration and foreign body left after procedure.

Satisfaction

I've been vocal that patient satisfaction is the necessary final leg on the barstool of health care reporting to round out the usual quality and cost data, and I was extremely glad to include the HCAHPS data in this report. Ten measures of patient experience after an inpatient stay populate this tab.

Services

The services tab shows the number of patients, the median length of stay and the undiscounted charge for a range of procedures and conditions, depending on whether you're viewing hospitals or surgery centers.

Staffing

The final tab shows two interesting proportion measures; the first is the ratio of full time nurses compared to contract staff, the second is the proportion of registered nurses, licensed practical nurses, and certified nurse assistants. I'd like to change this up to a pie chart next quarter. Also, on this page you can see the kind of beds this facility is authorised to operate.

Next Up

We're immediately looking at the nurse staffing data to get some more granular numbers based on various service areas, we'd like to add more procedures and conditions, and maybe bring in the rest of the process data. Regardless, Illinois hosted over 10,000 visitors today so there's no shortage of interest in these data, and we look forward to growing the site over the next year. First though, whynotthebest 2.0 - it's a busy season.

It's 106 hours to Monday, we got a full tank of coffee, half a pack of cigarettes, it's dark and we're wearing sunglasses. WhyNotTheBest.org will be updated in a few weeks with the most data I've ever worked with in one fell swoop, and will feature the most flexible search engine into our millions of rows of quality performance data yet.

‎ Full story...

Is it a Bird? A Plane? No, it's Electronic Clinical Data!

Faster than a speeding bullet, the clinical data portal we launched in June has scored it's first physician! It's pretty exciting to have built it, even more so to actually see a physician using it.

The portal allows a physician - or a group of physicians - to upload clinical data relevant to a given condition, in this case Diabetes. We ask for a set of specific, discrete data, not the entire record, such as blood sugar levels, blood pressure, eye exams and the like, and then using the data from the record we score the physician(s) against an evidence-based measure set that scores the doctor on his or her performance for that condition.

So, in this case, the doctor voluntarily submits to being scored for his performance treating and managing his diabetic patients. Any eligible physician who scores well on the measure will start receiving incentive payments from participating payors and purchasers, as well as be recognised in HealthGrades.com

Pretty cool, huh?

In addition, using the same engine we also scored two batches of physicians who are using the NextGen EHR package, thereby scoring their performance directly from the data.

As EHR adoption in the United States grows I anticipate more and more data being available to more and more performance assessors like ourselves, and I look forward to the next five years. Next step? Live report cards that tell you not how a hospital did last year, but how they did YESTERDAY.

Can't wait.‎ Full story...

Physician Performance Assessment

A bit of a departure from my usual transparency efforts, but this week we launched our Clinical Data Portal, a Web-based portal that allows physicians to upload clinical data directly from the chart or from an electronic medical record software and be assessed instantly on their clinical performance. Physicians who score high enough automatically start receiving bonus checks from the Bridges to Excellence program. Pretty neat stuff.

We have started with Diabetes and Cardiac Care, but we'll be rapidly adding more topics for physicians to measure themselves with. Physicians are free to become recognised for any of the topics, but if they fall under a BTE program area, they are eligible for real reward dollars. Next up we'll be processing batch files of data from EMR vendors. Exciting stuff.‎ Full story...

Why Not The Best? A New Web Site for Quality Improvement Professionals

Over the past six years I've been building consumer-oriented report health care cards pretty much non-stop. The field is young and short on evidence, but it's interesting work with noticeable impact. However, the truth is that the largest portion of readers of these report cards are not consumers, but providers.

To that end, I have long been wanting to build a report card for providers, not only to report their data but also to link directly to freely available improvement tools and knowledge directly from the report. My goal being, not only identify poor performance but right there, in the data, link to relevant interventions to impact that data in a positive way.

Well, it's done.



I met with the Commonwealth Fund earlier this year and they were looking to build something around their Why Not The Best? National Scorecard on U.S. Health System Performance. The Fund was looking for a vehicle to deliver their high performer cases studies, wherein hospitals identified as being high performers in a given set of measures were interviewed and the resultant wisdom distilled into case studies for other hospitals to learn from.

After many, many sleepless nights, I'm hugely proud to announce the launch of www.WhyNotTheBest.org - a quality improvement resource for providers from the Commonwealth Fund. A ton of very talented people have been working on this and we are all very, very excited to see how the site is received.



The site includes all clinical quality and patient satisfaction data ever released on CMS' Hospital Compare. This means 24 clinical measures and 10 satisfaction measures. In addition, we created 5 summary scores, 4 being topical and one overall composite quality measure.

Users can register for a profile that will remember all their choices for the next visit. This is important as there are many ways to personalise this site.

Profiles are a basic three step process, you choose hospitals that you wish to compare yourself to, then choose measures you wish to track, then choose benchmarks of care you wish to compare against.

We have every hospital in the US that bills Medicare. This is about 4500 hospitals. We have collected the quality and satisfaction data since the launch of the public dataset.

Using this data, we can calculate state and national averages, as well as top percentiles. Commonwealth then asked us to identify just those hospitals that reported all measures and had at least 30 patients in each of four topics, and using that subset we further stratified the population to ascertain the high performing hospitals.

This means that for any measure, you can load the current top 1% of hospitals for each measure.

Choosing A Hospital

You can choose a hospital to add to your profile in three ways. First, you can simply start typing a name and the database will let you know if we have a match.



Another interesting way to build a group of hospitals would be to find facilities that are like your own. Users can choose from a variety of characteristics to whittle down the list to a useful set of comparator hospitals. In the example below I've asked for all teaching hospitals in New York with 400 or more beds.



Finally, users can simply browse a map and grab hospitals of interest.



We don't limit the number of hospitals you can save to your profile, but if you add more than 50 the site will start to look a bit busy. Theoretically, you could store all hospitals in the country, but don't call me when it takes half an hour to load.

You can then choose any or all of the measures and benchmarks, these will then be available to you in any charts you choose to generate, and this is where things get interesting.

User-Generated Charts

We have implemented a dynamic charting application that will build charts on the fly based on your choices and your profile.

This means that for any chart you choose to load, it will populate dynamically based on the hospitals, measures and benchmarks you selected.

Here are a few examples:





Pretty nifty huh?

You can see in the example that we even calculated each hospital's rank in the nation for a given measure.

Each chart is followed by an array of options:



Among other things, you can download a comma separated file of the data used to create the chart, save the chart as an image to use in Powerpoint or similar, and click through to browse related case studies and intervention tools.



There are a wealth of additional features, please stop by and kick the tires! The site is www.WhyNotTheBest.org - enjoy! And please feel free to leave me your comments.
‎ Full story...

2008 Health Accountability Health Care Report

So here's why I haven't been posting too much lately...

I've been working on two major projects and this is one of them. I finally got the Health Accountability Report Card out today, it covers all hospitals and HMOs in NY, NJ, CT, VT and RI.

It's huge, but my other project - due in December - is even more exciting, stay tuned.

The report card started off as a printed HMO report in 1999, I first put it online in 2002 I think.

I've been slowly adding more and more to it, three years ago I added hospitals and last year I included NJ. Here's a brief, OK, not brief summary of what's in it.

The team that puts this together is large, all in all about 25 people contributed their expertise to the report. The short story is that it's a public data gumbo, it collects up about 20 data sources from which we extract roughly 250 measures. For each measure, if we can find a standard comparator we run a variance analysis to find out if the HMO or hospital deviates significantly from the state average. We then simply paste a red, yellow or green circle onto the data to give you a really quick idea of just how well the HMO or hospital you're looking at is doing.

On the HMO side you can review quality of care data from NCQA HEDIS scores. This covers stuff like getting the correct medicines, asthma and diabetes care, vaccinations and that sort of thing. Also for HMOs we have a customer satisfaction survey results from CAHPS surveys. We then scoop up retail premium rates for standardised plans.

Unfortunately states differ slightly in how they measure HMOs, so we can't make a national comparison across the board, so in this report you're seeing either a comparison to a state average or a regional average.

On the hospital side it gets way more complicated. HHS puts out the Hospital Compare data for download, so we grab that and crunch some additional measures to derive composite topical scores from the appropriateness of care measures. This year we also got to add patient satisfaction score and Medicare reimbursement rates.

The real fun part was requesting discharge data from five states. I actually requested data from every state that borders New York, plus Canada, and I went with the four I got back. That data is a full set of every single hospital discharge in each state, deidentified. In it you can see primary diagnoses, age, gender, enough to make some serious measurements. So, we have the AHRQ Inpatient Quality Indicators as well as new this year some Patient Safety Indicators, recently endorsed by the National Quality Forum. These include mortality rates and adverse events.

I did have to lose Caesarean section rates, but I'll keep hunting for an NQF endorsed measure for this.

In addition, we calculate number of cases, average length of stay and average hospital charges. Besides that there's the now-recurring Leapfrog patient safety measures. Cardiac surgery mortality came out this year for the most part, but it'll be back next year.

Here's a full list of all the measures.

Go check out the report and please, please, please send me your feedback.

Now I'm going to sleep.‎ Full story...

USA Hospital Patient Satisfaction Report

It's not widely known, but Medicare began collecting survey responses recently that asked patients a number of questions about their recent hospital stay. CMS has not restricted this to Medicare patients only, they call anyone who is eligible, regardless of payer. My boss was kind enough to give me the opportunity to take the data from Medicare and make a consumer-oriented report card that is easy and quick to use.

The site is AboutHealthSatisfaction.org and is completely free to use, no registration required, and does not contain ads. The source data is available on the Hospital Compare Web site put out by CMS, but it's hard to get to, hard to read, and restricts you to only a handful of hospitals to compare at a time. However, CMS makes the data freely available for download, so this is what I came up with.

Please go check it out, and let me know your thoughts and feedback here. I plan on updating the site every three months when the fresh data comes out. Right now I have three editions of the data, so you can even track numbers from quarter to quarter. Go have a look, and check back here to let me know what you think.‎ Full story...

New York Regional Health Care Report Card Goes Live

I know, I've been quiet. It's been a while since I posted, and here's why:

First and foremost, I have been buried in data all summer, getting this report card out. More on that in a minute.

Secondly, I determined to take some time to myself this year, so I've taken a slew of short camping trips to spend time with my son, do some fishing, sleep in a tent and generally get some of the city out of me.

Thirdly, I finally had my appendix out after my third and final bout with appendicitis.

So, now it's out. It's online at http://www.nyshaf.org/hcrc and I'd like to take this post to explain how I think it can be used, what purpose it serves, and talk about some of the data involved.

The report card is split into two main sections, HMO and Hospital. In both sections the report is split up by groups of measures that can viewed side by side, some quality, some patient experience, some cost.

One of the main problems with public report cards is that the health data is so complex and so arcane it's practically useless to the lay consumer. Sometimes a lower score is better. Sometimes a higher score is better. Sometimes the score is a percentage, sometimes a range. Sometimes it's a vague statistic definition, such as "not significantly worse that the national mean".

Another big problem is the tendency to predetermine what "good" is, and then try to communicate that top the viewer. Health care providers may say that high quality is paramount, cost is irrelevant. An HMO may say that the best care is given by better high quality, low cost providers. Some people think high volume is a good thing, other's don't. The list, and the permutations are endless.

Finally, the overarching goal is to present as many of these public reports as possible in a single space in a single format. For this year's report we've combined 25 publicly available reports which you'd normally have to obtain from 25 different sources.

So I set about trying to determine how to take all these kinds of different data and represent them in a singular fashion that would be easily and quickly comprehensible, without implying a judgement or a ranking. More than one reporter asked me "which hospital is the best?" My answer is always another question: best at what?

One hospital may be the world's best for cancer treatment; is that where I should go for a scheduled C-section?

There is no best hospital, there is no best doctor, no best surgeon, no best HMO. But there *are* providers that are better for you and your individual needs than others.

So, goals set: translate complex data analysis into visual nuggets of qualitative information, and allow the user to narrow down the qualities and measures that matter to them.


What you see above is one category of HMO data. More detailed information is available with a click or two, but at a glance you can see that for customer service one HMO stands out as being highly rated, three are more or less average, and one has some issues.

At least, I hope you can see that. If not, I just wasted a year.

So how can a consumer use this report card?

Example 1: Heart Attack

I have never heard of anyone having a heart attack and stopping to consult report cards on where they think they should go.

However, once you're in the hospital, your loved ones can consult the report to see how well the facility you ended up at is faring at providing quality treatment. If anything stands out as worrisome, bring it up with your doctor.

Further, several of the measures apply to the care you receive after your emergency treatment. Medications prescribed at discharge for example, including aspirin, as well as smoking cessation counselling.

Mortality rates for heart attack are relatively flat, everyone is going to be pretty much "as expected". If your hospital shows up red for mortality, bring this up with the hospital staff.

You may be looking at a co-pay or even worse, you're uninsured. Use the average charges bill to get an idea of how much the stay will cost you or your employer. You can also see the average length of stay for a heart attack patient, and if it's shorter or longer than you think is right, raise this with your physician.

The overall point is that the level and standard of care can and will vary from hospital to hospital, and nowhere is perfect. It pays to be informed, and remember you are your best advocate.

Example 2: Choosing an HMO

Some employers will offer you a choice of two HMOs. If you or your partner has an ongoing condition such as asthma, have a look at which of the two HMOs available seem to be more on the ball for that condition. Then check out the customer satisfaction ratings for that HMO.

If you don't have a choice and your HMO plan scores poorly, then use that as a reminder to make sure you get the care you need by advocating for yourself. Have a look at the care provided by other HMOs and demand that same care from your primary physician.

If you have children, compare the two youth sections. In my county, only one HMO has all green for appropriate youth care.

Example 3: Appendicitis

I have a bit of personal experience here. I had the symptoms of acute appendicitis for a few days, and finally went to the ER after the symptoms didn't go away. i was a little atypical, and was in little danger of perforation, but nonetheless you may be able to do what I did.

I looked at the five hospitals nearest me, and compared the length of stay for an appendectomy, the surgical infection prevention score,a dn the price.

I'm price-sensitive even though I have good health insurance. The company where I work pays for the health care I receive, and the last two years have seen a steady increase in both the company cost as well as my family rate contribution to the plan. Further, all my co-pays went up significantly this last year.

Because health care is so often paid for with invisible dollars it can feel a bit like Monopoly money, you never see it, you never worry about it, you get a long, arcane bill that you never have to worry about. However, I know that if I can get an appendectomy at hospital A for $10,000 less than hospital B, I just saved the company money. Saving the company money puts more dollars in the kitty for my next pay raise. And being relatively young and by all accounts rather stoic, I'd rather have the cash up front than ploughed into a pool for a rapidly greying employee population.

In my county, appendectomy hospital bills range from $7,000 to $30,000.

Let me repeat that.

In my county, appendectomy hospital bills range from $7,000 to $30,000.

That's before the surgeon's and anaesthetist's bills.

I actually placed more importance on finding a hospital that had a low average length of stay with good surgical infection scores. I wanted somewhere under $15,000 with three or so days stay on average.

The hospital I chose actually had a poor score for timely antibiotics, so instead of choosing a different hospital, I questioned the staff when my first antibiotics were given and ensured I was less than an hour from surgery.

Example 4: Mother to be

Some women search for hospitals that are less inclined to perform a C-section. In my county, the C-section rate varies from 17% to 34%. That's double the rate. For primary caesareans, the rate goes as low as 11% at a hospital that also shows a high rate of vaginal births after C-sections, clearly a hospital that can help you have a natural birth.

Others may choose the scheduled C-section route, and look for a hospital that performs a good number of the procedure. You can compare the number of times the procedure was performed in a year at all your local hospitals, as well as the surgical infection prevention scores.

Looking to get home quickly after the surgery? Some hospitals have you stay for three days, others for five.

--

In my next post I'll go into more detail about the different data sets including the different kinds of measures available, where we get the data from, and how accurate that data is.

In the meantime, I am very interested in hearing from anyone who has used a public report card similar to this one in a real life situation. Was it useful? Did it complicate things? Did you bring anything up with your doctor?

Finally, if you wish you had similar resources in your community, consult www.abouthealthtransparency.org/ and click your state. You can also check out www.consumerhealthratings.com/
‎ Full story...