Unwarping the playing field

Easing comfortably into my role as a "blogger," I've realized how easy it is to adopt a few key platform issues that tend to drive one a bit mad.  This blog isn't so fond of CAUTI, contact precautions, or devices that blow moist, warm particles over sterile fields (but then again, who is?).  We love diagnostic stewardship, influenza vaccination (um, mostly), and fecal transplantation.

Add advocating for better risk-adjustment of publically-reported HAI performance to my list.  A few months ago, I blogged about this issue and poor reporting validation by CMS -- now some excellent papers on improving risk adjustment have emerged, both from many FOB (friends of the blog) with senior authorship by Anthony Harris and his group at Maryland.  One focuses on SSI and one on CLABSI.  Their methodology is very similar and has some key features:

• They used comorbid conditions that are components of the Charlson and Elixhauser comorbidity indices
• These conditions were captured by diagnostic discharge coding that are currently routinely collected and submitted to CMS, limiting the data collection burden 
• They used conditions identified using Delphi consensus methodology from a survey of ID and infection prevention experts, providing some clinical credibility to the process

The authors examined the model performance and assessed changes in hospital rankings when compared to the traditional NHSN models. For SSI, a model containing procedure type, patient age, race, smoking history, diabetes, liver disease, obesity, renal failure and malnutrition showed good discrimination, and 86% of hospitals changed ranks within the cohort when the risk-adjusted model was used -- with 4 hospitals changing by >10 ranking spots.  For CLABSIs within the ICU, a model using coagulopathy, paralysis, renal failure, malnutrition and age showed improved predictability when compared to the NHSN ICU model, and 45% of hospitals changed ranking.  The authors note the clear limitations, including some of the challenges with using coded data, but these papers are important advances in the area of publically-reported HAI data.

You can read some of my more detailed thoughts in the accompanying editorial for the CLABSI paper (shameless plug).  At a time when there are many consequences for a hospital's performance on these surveillance metrics (e.g. last year my hospital received a draft quality incentive contract from a private insurer that required our large, tertiary care center to have ZERO of the Big 6 reported HAIs, to the tune of several million dollars in incentives), leveling the playing field to adjust for those factors that lead to HAIs that are beyond the control of the hospital is essential.  Thankfully, the CDC and HICPAC have recently chartered a new NHSN work group (disclaimer: Hilary and I serve on this group) and the issue of improved risk adjustment seems to be a major emphasis - fingers crossed that the field will start to level soon.

Are "One-Offs" Becoming Routine

one–off

play

adjective \ˌwən-ˈȯf\

After eight years investigating hospital outbreaks, and about 15 years trying to make the best possible use of surveillance data while at CDC, I still struggle with the tensions inherent in mixing surveillance and performance measurement. The past decade has been a roller-coaster of thrills and perhaps some spills in terms of attention, resources, refinement, and usefulness of HAI surveillance led by CDC; yes, you could probably blame me for several aspects of NHSN reporting you may find unsatisfying (take your pick – perhaps I will expand another day). However, I having recently retired from CDC and am transitioning to Emory Healthcare and Emory University. Although It has been almost eight months. It has been a fascinating transition. The learning curve is steep, and not just for re-entering clinical medicine (that is another story), but also navigating the pathway which integrates the business of healthcare delivery, quality of healthcare delivery, and research opportunities. Slightly easier was learning how to navigate the Emory Parking situation (took 3 months). Much easier was recognizing that the performance quality metrics linked to HAI prevention are getting a lot of attention and a lot of action. It only took a few sessions listening to the quality improvement teams reporting on their target HAIs to understand two things. First, the C suite leaders really care. I had assumed this while at CDC, but it was illuminating to see up close how hard these teams worked to influence HAI prevention. Second, it was becoming somewhat routine to report out on “exceptions to the rules” of HAI reporting. There are many names for those scenarios when an HAI is justifiably reported, but either considered not preventable with evidence based prevention practices or not clinically the infectious event represented by the HAI. While at CDC we routinely heard about these: CLABSIs that “shouldn’t really be counted”, MRSA BSIs that really “weren’t ours”, CAUTIs that really don’t represent an infection. Now these reported HAIs were being called “one-offs.”

The NYT reports the term ”one-off” comes from earlier industrial beginnings with the quantity of items produced in manufacturing process, such as taking one-off, two-off, or twelve-off the line to sample or give-away. However nowadays it can refer to any exception of the rule – such as a recent one-off boxing match that really should not ever have happened. 

In an HAI paradigm where we aim for 0 infections, one-offs may either be unavoidable (not preventable) or wrongly attributed to the device, location, procedure. I have historically known of these in terms of byproduct of using proxy measures. I had previously published an editorial on the value of proxy measures of infection as a tool for quality improvement (Meaningful measure of performance: A foundation built on valid, reproducible findings from surveillance of health care-associated infections).

In that editorial, we outlined necessary steps to reduce the inaccuracies inherent in using such an approach. Now that progress has been made in HAI prevention since 2010/2012, many of these HAI events that conspicuously remain and continue to plague our patients, often don’t fit neatly into the intent of the surveillance definitions. Left with these “one-offs,” it is often difficult to know what to do more to prevent them. Surgical patients with fistulas and central lines that don’t have an infection related to insertion or maintenance processes, neutropenic patients that don’t quite meet the definition of MBI-BSI, I have even heard of tissue transplantation related bacteremia categorized as CLABSI. No doubt, changes have occurred since 2011 to improve CAUTI reporting, and neutropenia-related bacteremia. However, the pace is slow. The one-offs are starting to pile up. Perhaps improved risk adjustment of HAI data will mitigate the influence of the one-offs on healthcare facility performance measures. Until then, kudos to the quality folks and infection control teams making prevention progress. However, I hope we can reward them soon with improved performance measures. Perhaps there are surveillance lessons that can be learned from these one-offs after all. 

If you are interested in sharing one-off stories I have started a registry here - maybe we can fill in some gaps and accelerate the process of changes in surveillance methods.

Effect of Nonpayment for Preventable Infections on SSI Rates Following Orthopedic Procedures

The debate regarding the effect of CMS's 2008 policy denying incremental payment for 8 complications of hospital care, also known as never events, is ongoing. Some studies have identified reductions in CLABSI and CAUTI after the policy was implemented, while others failed to find reductions in CLABSI, CAUTI or VAP, using a different dataset and methodology.  The impact of the nonpayment policy change on orthopedic procedures is less well studied.

Authors Jereen Kwong and colleagues published a study in the July ICHE that examined the effect of the policy on SSI following spine fusion, shoulder and elbow arthrodesis and repair, and spinal refusion procedures, but not hip and knee replacements (since they aren't covered under the policy). To determine the impact, they analyzed 20% of all inpatient discharges among patients ages 60 to 80yo using the HCUP National Inpatient Sample from 2000 to 2013. This is an all-payer sample in that it covers patients where Medicare is the primary payer but also non-Medicare populations.

The authors' primary hypothesis was that if the CMS policy had an impact, we would see larger reductions in SSI rates among Medicare vs control, non-Medicare patients. You can probably sense my concern immediately. How would bundles implemented to reduce SSI only have an effect on Medicare patients? It is unlikely that infection preventionists or clinicians would even have knowledge of their payor status. If non-Medicare controls have had the same exposure, they can't be used as controls!

Without getting into the methods details, the authors were expecting larger changes in SSI rates in the Medicare vs non-Medicare populations and they didn't find significant differences RR=0.9, 95% CI 0.8-1.1)  Looking at their Figure 1 below, we can see that SSI rates were slightly higher in Medicare populations (dashed line) vs non-Medicare populations before and after the non-payment policy went into effect. If I were to do this analysis, I would have looked for a change in intercept and/or slope in the SSI rates before vs after the policy combining both payor populations. Looking at the figure, there doesn't seem to be much change after the policy went into effect in either group. So we can agree with the authors that the policy had limited or no impact on SSI rates, but not for the reasons the authors state.

On the positive side, SSI rates have been declining over the entire 14-year study period - so surgeons and infection prevention folks can pat themselves on the back. Keep on doing whatever you've been doing!

The Playing Field is Still Warped . . .

Over the past decade, public reporting of facility-specific HAI performance has had a dramatic impact on infection prevention programs.  The increased awareness of HAI prevention by an increasing diversity of stakeholders (administrators, payors, patients, etc.) has arguably led to increased emphasis on HAI prevention and, in many cases, more resources to infection prevention programs.  These changes were very apparent at my institution where we increased our IP staff from 3 infection preventionists, 2 hospital epidemiologists and one administrative assistant to 9 IPs, 2 data analysts, 1 chart abstractor, 1 program coordinator, and partial support for 4 HEs (not including the antibiotic stewardship program support).  HAI performance is front and center among our annual quality goals, tied to department chair incentives, and the days since the last HAI are posted publically on our inpatient units for all to see.  Most importantly, our frontline healthcare workers understand and routinely discuss what were once surveillance acronyms like "CLABSI" and "CAUTI."  We've seen remarkable reductions in HAIs (and, more importantly, the associated patient harm) during this time (e.g. CLABSI in our ICUs have reduced 80% in the past 8 years). 

Our story is not unique.  Many hospitals have noted marked reductions in HAI rates.  One could argue that many, if not all, of the "low" and even "middle" hanging fruits have been tackled, and we are starting to reach the area where uncontrollable differences in patient risk factors/case mix may lead to different HAI performance across facilities.  With HAI performance tied to increasing financial consequences, however, the need to better insure a level and fair playing field across facilities is growing.  In this context, the recent report from the HHS Office of Inspector General on the CMS HAI reporting is very interesting.  The report focused on the validation of reported HAI data.  They found that while sufficient data were validated (as per regulatory requirements) and 99% of reviewed hospitals passed validation (only 6 failed), concerns were raised regarding how hospitals were selected for validation:

"However, CMS’s approach to selecting hospitals for validation for payment year 2016 made it less likely to identify gaming of quality reporting (i.e., hospitals’ manipulating data to improve their scores). CMS did not include any hospitals in its targeted sample on the basis of their having aberrant data patterns. Targeting hospitals with aberrant patterns for further review could help identify inaccurate reporting and protect the integrity of programs that make quality-based payment adjustments."

Gaming strategies that may be employed include overculturing (to designate an infection as POA), underculturing (if no blood cultures are collected . . . voila! No CLABSIs!), and adjudication/clinician veto ("I know that met the definition for SSI, but it was just a seroma . . . that I treated with antibiotics . . . uh, prophylactically . . . yeah, that's it!).  We have no clue how widespread these practices may be, but the OIG report notes that the current validation strategy should be enhanced to better capture gaming.  With the growing financial consequences placed on HAI prevention, it is paramount that everyone plays fair to better level the playing field.  Now, if we could also get more patient risk factors into the SIR models . . .

Should we track sepsis trends using administrative data?

A: Maybe, but only with extreme caution after adjusting for the influence of policy changes

One of the topics that our group at the University of Iowa is very interested in is the use of administrative databases for tracking the epidemiology of healthcare-associated infections, including those caused by resistant pathogens. Marin Schweitzer reported (open access) in 2011 that the ICD-9 code V09, which was commonly used for tracking MRSA trends, was a poor predictor of actual MRSA infection. More recently, Michi Goto completed a systematic review (free full text) that determined the accuracy of administrative data for surveillance of CAUTI, Clostridium difficile infection (CDI), CLABSI, VAP/VAE, postprocedure pneumonia, MRSA, and surgical site infections (SSIs). He found that administrative data detected CDI and SSI with moderate sensitivity and high specificity. For all other conditions, ICD-9 based algorithms had limited accuracy and utility.

Which brings us to a condition that we haven't much explored - sepsis. Sepsis has been in the news lately, since the Third International Consensus Definitions of Sepsis and Septic Shock (Sepsis-3) were published in February. I've included an image of a monument to a different Third International above for historical purposes. We would expect that these new definition might alter the incidence of sepsis as measured using administrative databases - but only time will tell.

Fortunately, Shruti Gohil and colleagues working with the CDC Prevention Epicenter program just published (free text in CID) an important evaluation of how previous changes in sepsis coding criteria, definitions and reimbursement might have altered sepsis rates. Using retrospective data (2000-2010) from California they determine the association between the release of CMS guidelines of sepsis coding in October 2003 along with the introduction in the MS-DRG reimbursement changes  for sepsis care in October 2007 and sepsis incidence. I've included two figures that showed (A) the change in incidence after each CMS change and (D) mortality per 1000 cases over the same period.

As appears obvious in the top figure, the marked increase in sepsis incidence, 3.6 fold over the study period, is temporally associated changes in CMS coding guidance and reimbursement. And fortunately, mortality appears to be decreasing. The authors suggest that some up-capture of less severely ill septic patients may be driving these findings. They also posit that background increases in sepsis, separate from those associated with CMS changes, could be driven by early recognition programs, such as Surviving Sepsis Campaign (2004), and improved diagnosis. The most important conclusion is this: "it is imperative that reported sepsis rates based on administrative data account for policy-related effects...interpretation of recent epidemiologic trends in sepsis based on administrative data should be approached with caution." 

Measurement fatigue: The backlash

Anyone who’s been in the hospital infection prevention business for any length of time is familiar with a specific form of cognitive dissonance. We believe, on the one hand, that the publicly-reported, metric-focused, pay-for-performance (PFP) environment has brought increased resources to infection prevention and resulted in a real decrease in healthcare-associated infections (HAIs); yet we believe, on the other hand, that these high stakes have led to a number of unintended adverse consequences, including gaming of HAI definitions, an unhealthy focus on measures that may not merit the resources and attention, and have engendered cynicism as it becomes apparent that PFP measures may not correlate with actual quality or value.

Over the past week, I’ve read three pieces that make me wonder if we’re reaching a tipping point, as clinicians begin to push back effectively against the proliferation of “measures”, “metrics”, “performance targets” (whatever you wish to call them), in an attempt to seek a balance between them and the words of Francis Peabody, that “the secret of the care of the patient is in caring for the patient.” I’ve pasted some key quotes below from each of these pieces.

First, an excellent opinion piece from Dr. Robert Wachter in the New York Times:

"All of this began innocently enough. But the measurement fad has spun out of control. There are so many different hospital ratings that more than 1,600 medical centers can now lay claim to being included on a “top 100,” “honor roll,” grade “A” or “best” hospitals list. Burnout rates for doctors top 50 percent, far higher than other professions. A 2013 study found that the electronic health record was a dominant culprit. Another 2013 study found that emergency room doctors clicked a mouse 4,000 times during a 10-hour shift. The computer systems have become the dark force behind quality measures….

....Our businesslike efforts to measure and improve quality are now blocking the altruism, indeed the love, that motivates people to enter the helping professions. While we’re figuring out how to get better, we need to tread more lightly in assessing the work of the professionals who practice in our most human and sacred fields."

Next, from a fascinating report of hospitalists unionizing in Oregon:

"To work in a hospital today is to be constantly preoccupied with money, and one of the more grating features as far as the Sacred Heart hospitalists are concerned has been the administration’s celebration of “skin in the game.” That means creating financial incentives for doctors to hit performance targets — like lowering patient’s length of stay and doing well on patient satisfaction surveys. The phrase entered the Sacred Heart lexicon in 2014, but the underlying concept has spread throughout the profession in recent years…. 

…the increasing focus on metrics like readmission rates and hospital-acquired infections had created more work for hospitalists, who are responsible for a lot of documentation."

And finally, from a NEJM piece on “Taylorism” in medicine:

"Instead of gaining happiness minutes, clinicians are increasingly experiencing dissatisfaction and burnout as they’re subjected to the time pressures of Taylorism and scientific management in the name of efficiency. We have watched colleagues fleeing to concierge practices, where they have control over their schedules. Others have taken early retirement, unwilling to compromise on what they believe is the time needed to deliver compassionate care. Some have moved into management or consulting positions, where they tell others how to practice while unburdening themselves of their clinical load. Just as Taylor enriched himself by consulting for companies, a growing and lucrative industry has emerged to generate and enforce metrics in medicine. By 2014, the Centers for Medicare and Medicaid Services alone had mandated the use of more than 1000 performance measures. As the Institute of Medicine recently reported, such metrics have proliferated, though many of them have little proven value."

(This last piece probably should have given a shout-out to a piece by Mike that started by similarly invoking Frederick Winslow Taylor.)

The real reason ID docs are the lowest paid physicians: AMA's Relative Value Scale Update Committee

In keeping with our annual tradition, Dan recently posted on the declining interest in ID fellowship slots filled through the match. Re-reading his 2013 and 2014 posts, it's quite clear things have not improved. It's also clear that IDSA takes the decline very seriously. In fact, two IDSA presidents, Dr. Stephen Calderwood and Dr. Johan Bakken, have taken time to post IDSA's diagnosis and responses to the public health problem. Both described the relatively poor compensation provided for ID services.

Quoting Dr Bakken: "There is no question in my mind that the financial student loan burden and inadequate reimbursement for ID services are major disincentives for young physicians contemplating a career in ID. IDSA alone does not have the power or means to rectify the problem, but we are working very hard with legislators and policy makers on Capitol Hill."

Without getting too much into the weeds, I wanted to share with you why I think Infectious Diseases is so poorly reimbursed compared to every other subspecialty. The reason is as old as politics - we have no representation on the AMA's Relative Value Scale Update Committee (RUC). Since 1991, CMS has collected advice from this AMA Committee on how much "physician work" is involved in delivering a particular service. This committee is important, since CMS agrees with the committee's recommendations almost 90% of the time. And as you can see in the figure I posted below, there is unequal and unfair representation on this committee. Some specialties are under-represented based on the number of services they provide (i.e. primary care) and certain medical subspecialties (e.g. nephrology, hematology) are only represented on a rotating basis while others (e.g. cardiology) have a permanent seat. Looking closely at the list of subspecialties, I don't see any Infectious Diseases representation!

So, if we want to fix ID, we need permanent representation on this committee. It is a complete travesty that the highly reimbursed procedure-focused subspecialties are fully represented but the "cognitive" subspecialties (endocrinology, ID, rheumatology) are invisible. IDSA needs to demand equal and fair representation.

SHEA 2015 Update: Abstract Deadline (January 30th), Pro-Con Session and Certificates

SHEA - Orlando, May 14-17, 2015

The SHEA spring meeting in Orlando is really coming together. With the abstract deadline approaching (January 30th) and the full agenda finalized, I wanted to highlight a few things that you should know as you rush to register before the February 13th early registration deadline.

1) Register before February 13th and save $100. If you aren't already a SHEA member - become a member at least 48 hours before registering for the meeting to save even more! Hope to see you in Orlando - May 14-17, 2015.

2) This year, there are two certificate courses that you can select when registering. In addition to the annual SHEA/CDC Training Certificate Course in Healthcare Epidemiology, there is a new SHEA Certificate Course in Post-Acute and Long-Term Care Track. When you register and attend either course, you will receive a certificate in addition to CME/CE. Specific sessions for the SHEA/CDC course (purple) and LTC course (orange) are in the grid below.

3) Finally, I wanted highlight the Friday afternoon (May-15th) Pro-Con session titled "Does Pay for Performance Reduce HAI?" In this session the "Con" side saying the policy doesn't work will be discussed by Grace Lee, MD MPH. As you know, she published an important study on the topic in the NEJM (2012). In that quasi-experimental study, she showed that there were no changes in CLABSI, CAUTI or VAP rates before/after the 2008 nonpayment policy implementation in 398 NHSN hospitals. You can read my blogpost on the study here. On the "Pro" side, Teresa Waters, PhD will discuss her recent JAMA-Internal Medicine study (2015) that showed that the CMS policy was associated with reduced HAI. Using a quasi-experimental design and data from 1381 US hospitals participating in the National Database of Nursing Quality Indicators (NDNQI), she showed that the same CMS policy was associated with an 11% reduction in CLABSI and a 10% reduction in CAUTI but no change in falls or pressure ulcers. So did the policy work or not? Gotta attend SHEA 2015 to find out!

2015 Agenda - Click to enlarge

What the HAC?!

The University of Wisconsin (UW) Hospital is an excellent institution with a stellar infection prevention program. In fact, last year they won the prestigious U.S. Department of Health and Human Services (HHS) Partnership in Prevention Award, which recognizes “prevention leaders in the U.S…who have achieved wide-scale reduction and progress toward elimination of targeted health care associated infections.” Sadly, this achievement may not be enough to keep HHS from levying financial penalties against UW for high infection rates. 

Yes, the Hospital Acquired Condition (HAC) scores have come out, and have been generating a fair bit of media coverage, focused on those hospitals most likely to face financial penalties. Funny thing, though—the hospitals most likely to lose money under this program share a lot of characteristics:

“who is getting penalized? Large, urban, public, teaching hospitals in the Northeast with lots of poor patients. Who is not getting penalized? Small, rural, for-profit hospitals in the South. Here are the data from the multivariable model: The chances that a large, urban, public, major teaching hospital that has lots of poor patients (i.e. top quartile of DSH Index) will get the HAC penalty? 62%. The chances that a small, rural, for-profit, non-teaching hospital in the south with very few poor patients will get the penalty? 9%.”

Interesting. Explanations for these findings include: (1) small size, rural location, southern region and for-profit status magically translate to higher-quality, safer care, or (2) this HAC metric is bullshit, as it obviously doesn’t adequately control for myriad variables that are associated with the score but that are not indicators of quality and safety. What variables? Intensity and accuracy of surveillance, and variation in infection risk of the different patient populations, for starters.

I can overlook bullshit when it brings more attention (and resources) to the critical task of infection prevention. Unfortunately, this particular form of bullshit does the opposite (unfairly punishing already cash-strapped hospitals with financial penalties). As others have pointed out, the current HAC metric is well-intentioned but obviously flawed, and in desperate need of fixing.  

Did the CMS no-payment rule impact hospital HAI prevention practice?

Beginning in October 2008 CMS stopped reimbursing hospitals for excess costs attributable to CLABSI or CAUTI.  While numerous studies and surveys have linked high compliance with HAI prevention bundles for CLABSI and VAP to reduced infection rates, few studies have looked at whether CMS no payment rules improved HAI prevention process measures.

To answer that question, Sarah Krein at the VA Ann Arbor Healthcare System completed surveys of VA and non-VA hospital HAI prevention practices in 2005 and again in 2009.  Their hypothesis was that if adoption of HAI bundles differed between non-VA and VA facilities, some of this difference could be do to the CMS no payment rules since VA facilities aren't directly affected by CMS rules.

The results are pretty interesting and don't really support any impact from the CMS no payment rules.  For CLABSI, both VA and non-VA hospitals reported significant increases in bundle component use with VA having higher use in both 2005 and 2009 (see graph below).

Similar results were reported for VAP and CAUTI.  The authors conclude by saying that "the CMS payment rule is likely not the primary driver of the increased use of infection prevention practices among US hospitals over the past several years."

Source: Krein et al. JGIM July 2012

Dr. Berwick goes to Washington

The Story is one of the podcasts I listen to while I walk to work. This week’s episode is well worth a listen—an excellent interview with Don Berwick about his time with CMS. I can’t comment much more without making this a political blog, which is not my intent. But I urge you to listen at least to the story of his meetings with two senators, starting at about 14:45 into the podcast (downloadable from the website or from iTunes).

High stakes and low rates

The recent scandal surrounding a college that submitted inflated SAT scores to improve their ranking in the US News & World Report should remind us of a simple fact—the higher the stakes, the more likely cheating will occur.

Now that CMS is publishing ICU CLABSI data on their hospital compare website, and using the data in payment formulas, the stakes for hospitals could hardly be higher. And available data suggests that many hospitals stray from strict application of NHSN definitions, reporting misleadingly low CLABSI rates. We’ll soon be publishing results from a survey we did of hundreds of infectious diseases clinicians involved in CLABSI reporting at their institutions. Given patient scenarios that clearly met the NHSN definition for primary CLABSI, what percent do you think responded that they would report them as primary CLABSIs? I can’t go into detail given that this is as-yet-unpublished data, but I suspect the results will surprise even those most critical of publicly-reported HAI data. So I’m dismayed when I read quotes like “several hospitals that serve the sickest patients have been able to achieve infection rates of zero for several years”. The statement is obviously false—but even if it were true, it simply means that the hospital isn’t applying NHSN definitions correctly.

Good riddance!

I’m so happy to see this flawed performance measure dropped. I think I’ll take a 5 minute break to write a blog post, then another 5 minute break to stare out the window and marvel that a measure with such toxic unintended consequences was actually abandoned!

h/t to Edward Goodman, via a post on the EIN this morning.

“Not your father’s CMS”

I just returned from my first HICPAC meeting….and no, I won’t risk being thrown off the committee for discussing all of our deliberations on this blog (at least right now I won’t). However, I did want to point out a fun moment from a presentation by John O’Brien from CMS. John gave an update on the CMS Partnership for Patients, which we blogged about earlier.

The PFP is still nascent, and we’ll see how much of an impact it has on the ground. But the presentations by Joe McCannon at SHEA and this one from John O’Brien were both encouraging, and almost certainly reflect the impact that Don Berwick has had on the culture there. The photo above was taken at the break after John’s lecture. As you can see, there is a painting on the back of his suit jacket, depicting patient safety themes. During his talk he briefly discussed this “walking art”, and how it helps start conversations in airports, at conferences, etc. The painting is the work of Regina Holliday, a Washington DC-based patient rights arts advocate. The full story and explanation of the images on the jacket can be found on her blog, here.