Reengineering Healthcare - The View from Other Industries

Alice: Oh, no, no. I was just wondering if you could help me find my way. 

Cheshire Cat: Well that depends on where you want to get to. 
Alice: Oh, it really doesn't matter, as long as... 
Cheshire Cat: Then it really doesn't matter which way you go. [1]

Some time ago, February 2008 to be exact, I published an MIT working paper entitled GDP-Based Productivity in Ambulatory Healthcare: A Comparison with Other Industry Segments, 1998-2005[2], wonderful title, I know… I looked at a set of productivity measures over the years 1998-2005 for ambulatory healthcare, auto manufacturing & the information industries, & what I found surprised me. Productivity measures such as value added per hours worked or gross-output per gross-input were generally better for ambulatory healthcare than for auto, & only slightly less productive than for information industries. Value added per total compensation was actually better in ambulatory healthcare than in the other benchmark industries, but that was mainly because compensation was comparatively low in healthcare. These measures are based on labor productivity that is the contribution of the direct work that people do to produce an output or product.

OK you say… that’s semi-interesting, but what does it have to do with re-engineering, healthcare reform or Alice’s feline adversary? Bear with me for a minute & I’ll try to explain. It does make sense (at least to me…). Over the past 25 years or so, economists have developed measures of productivity that are not based on direct labor. The main one of these is called total factor or multifactor productivity (MFP)[3] & it measures the effect of non-labor, usually capital-based, inputs on outputs. I’m already way to deep in the weeds here, but suffice to say that multifactor productivity is taken as a measure of how well industries use capital, & is thought to reflect the efficiency of “ changes in characteristics and efforts of the work-force, changes in managerial skills, changes in the organization of production, changes in the allocation of resources between sectors, the direct and indirect effects of R&D, and new technologies”. More specifically, effective inputs are: improvements in workflow, more effective R&D investment, adoption & usage of new technology. The Bureau of Labor Statistics recently published multifactor productivity statistics for industries up until 2011[4], the last full year that stats are available. An analysis of these stats for the 11 years 2001-2011 has the following summary:

	Auto	Information	Amb. Healthcare
Multifactor Productivity[5]	98.4	102.8	99.9
Standard Deviation	3.83	3.95	1.26

The key facts here are that MFP in information industries is a good deal higher than in Auto or Ambulatory healthcare, that MFP is higher in Ambulatory Healthcare than in Auto & that there is almost no variation in the MFP measure in Ambulatory Healthcare over the 11 years.

 Information Industries

Ambulatory Healthcare

Auto Manufacturing

Auto had a significant (0.99) decrease in MFP starting with the beginning of the recession in 2008 & had still not recovered as of 2011, while Information Industries had a significant (0.99) increase during the same time. MFP in Ambulatory Healthcare stayed relatively constant for the period. The explanation that is generally accepted for this pattern is that manufacturing process in Auto went through a period of intense examination & redesign in the 1990s & as did the Information Industries a bit later (1995-2005). This effort resulted in improvements in MFP, particularly from workflow redesign, adoption & use of new technologies & efficiency of R&D investment. The loss of productivity in Auto was associated with major losses in sales, bankruptcies, loss of employees etc. in the 2007-2010 period.

Which brings us to re-engineering in healthcare. In 1990, Michael Hammer, then at MIT, published an article in the Harvard Business Review titled: Reengineering Work: Don’t Automate, Obliterate[6]. His main points were that most work being done did not add value for customers & that automating this work only perpetuated the inefficiencies in more effective forms. He suggested that what should be done is abandon ineffective work & redesign work process to be in line with corporate goals & customer needs. This effort was called reengineering, & many authors took up the debate about through out the 1990s, just as many corporations undertook to reengineer their way of working.

The increases in MFP in the Auto & Information Industries, & even the subsequent decrease in Auto are evidence of their reengineering efforts, especially in workflow redesign & use of new information) technologies. As I said in my ESD Working Paper (op. cit.), “The ambulatory health segment has not made many of these changes up until this time. …investment in new technologies, whether in hardware (medical devices and computers) or software applications has not been a priority. Neither has structural or management reorganization. This segment is only just now faced with needing to make some of these changes as it focuses on the adoption of electronic healthcare records (EHR), which requires investments in hardware and work reorganization. Ambulatory healthcare has made progress in labor productivity without making these changes, but not in capital efficiency. In order to consolidate and extend the labor productivity gains made from 1998 to 2005, and to begin to make gains in capital efficiency, this segment will have to address the key factors that affect these productivity changes. These include increased research and development spending, increased technology adoption and addressing economies of scale as well as making serious efforts at productive structural and work reorganization, and improving managerial skill. Advances in these areas will allow ambulatory healthcare to lower costs and improve both labor productivity and capital efficiency.”

Translated from academize, this means that Ambulatory Healthcare has yet to seriously address those areas that will allow it to make progress, namely: clinical & operational workflow redesign, adoption of new information technologies (other than EHR) & more effective use of R&D investments. The Federal Meaningful Use program has ensured that healthcare organizations adopt EHR technology, & this has required a good deal of workflow change, but as we have seen in several of my previous posts, adopting EHR technologies & qualifying for meaningful use reimbursement does mean that an organization is using technology in a (clinically or operationally) meaningful way, or that the workflow changes made are effective in terms of any goals other than to be able to use a specific vendor’s EHR. This is the point of the Alice quote at the beginning of this piece. The Cheshire Cat is quite correct – you can’t get where you want to go unless you know where that is. If all you want to do is to qualify for Meaningful Use, than what has already been done is adequate, but if what you want to do is really address the Triple Aim, you have to do much more. How would you align workflows, especially clinical workflows, as well as the adoption of newer health information technology along with the more effective use of R&D investment if you really wanted to focus on the Triple Aim. Here are some suggestions:

•      The ideas behind meaningful use, Patient-Centered Medical Home, Accountable Care Organizations etc. are correct in the sense of a high-level direction, even of the specific details of the programs are less than effective in actual improvement of clinical outcomes & achievement of cost reductions. The first thing to do (IMHO) is to evaluate what changes would actually be effective in addressing the Triple Aim, & then to move toward making those changes. Some of these changes appear to be:
• Reengineering both administrative & clinical workflows to emphasize continuity of care, care transitions, medication reconciliation & team-based care planning & treatment where the patient (&/or caregiver) is part of the team.
• One of the first “technical” changes would be the provision of a complete, integrated patient record. This would include primary care, behavioral health, dental, patient supplied data & potentially data from external sources (such as public social media)[7]. Such a record is essential for realistic care planning & treatment. Current EHRs do not provide this type of integrated record & are currently under no regulatory or commercial pressure to do so. Many organizations are attempting to do this level of integration either themselves or through a cooperative vendor. This situation has to change so that integrated patient records a the general case, not the exception
• Of course, this kind of change is not entirely technological. One of the main reasons that such records is cultural, both on the healthcare & the vendor side. Vendors have to be rewarded commercially for providing the capability to produce such a record & providers have to be reimbursed for using one in an integrated & patient-centered manner.
• Workflows have to be restructured so that teams are enabled & so that patient input is expected (& welcomed). Auto, Information Industries & other industries (such as Aerospace as well as continuous process industries such as chemical manufacturing) went through years of process redesign, mostly by trail & error (enabled or not by process consultants). Ambulatory healthcare has only just started this process with respect to meaningful use of EHRs. That effort has to be extended to include the provider, care team & patient addressing all aspects of the triple aim.
• This reengineering will necessarily mean that EHRs & other HIT will need to be redesigned (fundamentally, not just their UIs) in order to enable the new integrated focus.

That’s at least a start…

Next up:

•         Part of the problem with reengineering in healthcare is that reengineering works well in segments that are actual markets (because of competitive & cost pressures), but less well in segments that are not. I (& many other people better qualified to know than me) believe that healthcare is not a market so these competitive & costs pressures are not effective at structuring how the segment functions. Why isn’t healthcare market? Stay tuned for that…
•     I spent almost three years working as the corporate technical leader for the Digital Equipment Corporation to General Motors. This was during a time (1986 & following) that GM was undertaking a huge reengineering effort in engineering & manufacturing called C4. I participated deeply in this effort & think there are some lessons learned for healthcare, very tough lessons. I’ll try to explain the relationship & what it might mean for healthcare reengineering.

---

[1] Alice in Wonderland. 1865. Charles L. Dodgson (Lewis Carroll). Woodblock Illustrations by John Tenniel for the first (1865) edition. Read in the 1993 edition: The Annotated Alice: Alice’s Adventures in Wonderland & Through the Looking Glass. 1993. Martin Gardner (ed.). Random House.

[2] Hartzband, D.J. 2008. GDP-Based Productivity in Ambulatory Healthcare: A Comparison with Other Industry Segment, 1998-2005. ESD-WP-2008-11. Engineering Systems Division. Massachusetts Institute of Technology

[3] http://www.bls.gov/mfp/. The Bureau of Labor Statistics MFP site

[4] http://www.bls.gov/mfp/mprdload.htm.

[5] 2005 taken as index year (100.00)

[6] Hammer, M., (1990). "Reengineering Work: Don't Automate, Obliterate", Harvard Business Review, July/August, pp. 104–112.

[7] See my posts of December 16 2013 & January 22 2014 (posttechnical.blogspot.com/)

Progress in Healthcare Reform - Part 3:HIE & ACO

My last two posts addressed progress in healthcare reform from the standpoint of meaningful use & PCMH. In part, I tried to critique some of the recent criticisms of these programs, but in truth, I agree with some of that criticism & laid out why. This is the last post in this series (but certainly not the last post on progress in healthcare reform), & its focus is HIEs & ACOs. The issues with HIEs were covered in a late January (2014) Black Book survey[1] reported in Government Health IT found that 95% of payers, 83% of hospitals & 75% of providers thought that publicly funded HIEs had flawed business models & provided no meaningful connectivity. Let’s look at each of these separately.

Public HIEs’ business (sustainability) models are dominated by the $546M in funding provided by the HITECH Act. As this funding ends (actually, it has ended), public HIEs have had to try to develop other means of sustaining their operations. Among these have been charging (large) fees to stakeholders, but HIE stakeholders are in very different positions relative to paying for services. Most provider organizations are not in the position of being able to pay large amounts for HIE capabilities. Many do not have the financial resources to do so, & those that do, like the big payer/provider organizations, do not want to pay for capabilities that have not been shown to be essential (yet?).

A second potential means of providing sustainability is the development & offering of new capabilities: IT & data management, data warehousing, analytics & nontechnical capabilities such as financial management services. Charging constituents for these services could provide financial stability, but first those services have to be developed to a point where an organization (that may already be doing much of this) will pay for them. This is not as easy as it seems.

The real bottom line, in my opinion, though, is that HIEs have yet to provide a service that is necessary. The original use cases that consisted of a patient serendipitously needing emergency or trauma care far from their primary healthcare provider just does not happen often enough to sustain a half billion dollar investment & less ambitious use cases are not compelling for providing needed care. If HIEs were to develop a model that provided information for care in situations that actually exist in the real world, there would not be a sustainability problem. I work quite a bit with Federally Qualified Health Centers (but the same example is mostly true of ambulatory care in general), & many people who get their ambulatory care at FQHCs get trauma-based (& a good deal of other care) care at hospital EDs. This is especially in urban environments. A person getting their care at one of the Ryan Community Health Network sites in Manhattan, NY could potentially get non-ambulatory care at any one of hundreds of hospital sites. They might have to choose from 10-25 hospital sites in walking distance. What happens when they walk into one of those sites? Chances are that the years of medical records from their health center, or even just their medication history, will not be available to the ED providers. This is a sustainability model right here, & it doesn’t have only to be in NYC.

Of course, the other issue with HIE is that they actually have to work – that is, they have to be able to share critical clinical & administrative data among constituents. This is only incompletely realized in most cases & will have to substantially improve before any sustainability model can succeed. The proof of this is the amount of vendor churn in HIE. I wrote about this as early as 2011[2], & my current observation is that the situation has not changed for the better with many HIEs looking to change their vendor. Much of this is connected to EHR churn as a 2013 survey[3] found that one in six medical practices were looking to change their EHR vendor in 2013 & one of the primary reasons was HIE support.

What can be done to improve this situation?

• First, vendors must provide the connectivity & interoperability for data sharing that users require in order to have the basic functions of an HIE work. Sustainability is a moot issue if the HIE is not providing adequate function.
• If HIEs are to provide additional services (beyond the connectivity & interoperability capabilities referenced just above), then they must provide those services that users want & need, & they must do this in a way that makes these services easy to understand & to use. Examples might include: 1) data warehousing including ETL & normalization so that the data is usable, 2) analytics related the warehousing including “canned” analyses for a “top 10” list of questions/issues, 3) support for accountable care, 4) outsourced IT management etc.
• Capabilities for more “local” data sharing aligned with healthcare usage of the patient population being served.

This brings us to ACOs. Section 3220 of the Affordable Care Act established the Medicare Shared Savings Program & CMS published a final rule on this program in November 2011[4]. ACOs are healthcare organizations that commit to share savings & risks for at least 5,000 Medicare fee-for-service patients. Simply put, the cost of care for these patients is compared to CMS benchmarks & if the organization’s cost is at least 2% less than the benchmark, they qualify to “share” in the savings through higher reimbursements. There is also a shared risk model which organizations opt into where if the savings are below benchmark level (i.e. no savings), the organization may be required to pay CMS back part of it’s reimbursement. The details are, well… substantially more detailed, but you can read the 190 page regulation yourself if you are interested.

So how is this program doing? If we look at the first year of the ACO Pioneer program we can see pluses & minuses. CMS selected 32 advanced healthcare organizations for the Pioneer Program. Of these, nine (28%) dropped out in the first year for various reasons. 13 organizations (41%) were paid bonuses for showing savings. CMS’ estimate for savings totaled $87.6M that is .015% of Medicare expenditure for 2013. Participants’ issues with the program, including those that dropped out, centered around administrative complexity & the relatively small amount of reimbursement compared to the effort of producing savings. If CMS’ expectation is that some form of this program will eventually replace fee-for-service payments, they have a very long way to go.

Why, then, are so many HIEs focused on morphing into ACOs? The answer seems pretty straightforward – sustainability. ACOs offer one of the few models for a healthcare organization that pays money for a level of care & cost savings that the organization would want to achieve in any case. The real question is: “Is this a good idea?” The answer would be yes if it were easy to provide “accountable care” to 5000+ Medicare patients in such a way so that their outcomes were improved & cost savings could be measured against a CMS benchmark. This does not seem to be the case, though. If some of the best healthcare organizations in the country are having trouble providing improved care with the required “shared savings”, then it will be difficult for less well-resourced organizations to do so.  An example would be Texas Health Resources (consisting of 13 hospitals, managing 41,000 Medicare beneficiaries in their ACO), a 2013 HIMSS Davies Award winner for EHR adoption. THR dropped out of the Pioneer Program as they failed to achieve shared savings & would have owed CMS $6M-$9M by the end of 2013[5]. As always, my question is what can be done to improve the situation, although I have to admit we are in very early days with ACOs & I have not worked much on this… however that never stopped me from making suggestions[6]:     

• Every aspect of the program has to be simplified. This includes how savings & benchmarks are calculated, eligibility requirements, administrative requirements, reporting,… every aspect.
• Some program details have to be rethought:
• The amount of reimbursement (shared savings returned) needs to be concomitant with the effort to create those savings. The best data we have on how this might work is from the Physician Group Practice demonstration[7] run by CMS from 2005-2010. If a new ACO made the same investment that the PGP demonstration groups did in their first year ($1.7M), they would need to make a 20% profit to break even over their first three year ACO contract. This seems unlikely.
• Patients in the ACO need to be able to be identified. It is very difficult to measure things like hospital readmissions (& many other quality measurements) if all you have are statistical characterizations. If ACOs are going to be about improvements in population health, then that’s what they should be measured on.
• CMS must do better in coordinating sharing of claims & other data so that participating organizations have the data they need to monitor quality & costs.
• A realistic care model must be part of the ACO model. We know from looking at HIEs that most patients get their care both within the HIE & outside of its participants. The same is true of ACOs, & so the shared savings must take this into account. ACOs cannot be held accountable for patient-driven costs that the organization is not involved in, but they are currently held responsible for overall quality of care & all costs. Conversely all relevant costs need to be taken into account if cost is going to actually be reduced. I don’t know what the answer to this one is, but it needs to be worked on.
• Part of the problem with the shared cost model is that savings are in part calculated on growth rates from a national sample. CMS’ own data[8] show that most ACOs have costs in line with their region & local areas, & that these may differ greatly from the national averages. Costs need to be evaluated based on local conditions, especially where those are different from national averages. 

So,… a lot of issues with HIEs & ACOs. Are these organizational & clinical models the future of healthcare delivery? CMS seems to want to evolve the ACO model into an eventual replacement for their Medicare fee-for-service model. Anything that realistically moves us away from fee-for-service models, & also improves outcomes while reducing costs is an essential & welcome change to the system. Are HIEs a step along this path? For all of the investment in HIEs, greater than $0.5B to date, I do think that they are a temporary step. HIEs need to continue to find sustainability models. I do not think that their governance model lends itself to profitable service provision (not what they were designed for) & the sharing of healthcare data over large distances also does not seem to be a sustainable model. One question to ask about HIE sustainability is: “Were these organizations meant to be self-sustaining businesses?” Obviously, their evolution to ACOs is one way to deal with the major issue facing them, but that then begs the question of whether the majority (or even a small number) of the approximately 250-280 HIEs in the country can successfully make this change, & if they do, can they succeed as an ACO? I think the answer is no unless CMS works with healthcare organizations, payers & all constituents to evolve the ACO model to be simpler, both to qualify for shared savings & to improve the incentives to do so.

Coming up:

•         Many people, myself included, have said that healthcare is really not a market. What does this mean in terms of our current reform effort? & what are its implications for the future healthcare system?
•     Other core industries have done substantial re-engineering work that made them much more effective. Has this work been done in healthcare? What would doing it look like?
•     To be supplanted by relevant & interesting topics as they come up… 

That should be enough for a while…

---

[1] http://www.prweb.com/releases/2014/01/prweb11503131.htm

[2] http://www.hiewatch.com/perspective/change-good

[3]http://www.questdiagnostics.com/dms/Documents/hit_quality_solutions/qs_vendor_enewsletter_oct2013.pdf

[4] 42 CFR Part 425 Medicare Shared Savings Program: Accountable Care Organizations. Federal Register 76(212).

[5] THR has said the two biggest impediments to achieving shared savings were lack of CMS to provide timely claims data (to monitor quality & cost) & lack of ability to coordinate care for managed patients seeking care outside of the ACO. Other organizations had these same problems. http://www.modernhealthcare.com/article/20130720/MAGAZINE/307209949/1138. Accessed 3 April 2014.

[6] At the Digital Equipment Corporation, we used to say that the (unofficial) Consulting Engineers’ motto was “Often wrong, never uncertain!”

[7] Health Affairs. Health Policy Brief. Next Steps for ACOs. Health Affairs Blog. January 31 2012.

[8] http://innovation.cms.gov/Files/reports/PioneerACOEvalReport1.pdf