Cost Shifting in Healthcare

press shift

There is a widely held belief, perhaps unspoken but no less strongly held, that the healthcare business is a zero-sum game.

Consider how healthcare dollars are generated.   A hospital system, care facility, or provider provides a service to the surrounding area, termed a catchment area.  Those covered lives in the catchment are expected to generate a certain amount of healthcare expenditures on an aggregate, population basis.  This is modeled by insurers and hospital systems for budgetary purposes.  Given the number of people in the catchment area, the age, socio-economic status, general degree of illness, and type of insurance, finance professionals and actuaries can make an estimate of expected healthcare dollars by payors (insurers, government) to providers and facilities on a per patient basis.

While modifiers, complications, and co-morbidities can alter the real billing for a particular patient and encounter, on aggregate most in the healthcare industry tend to think that these care dollars will either be captured by their system or a competitor.  Hence, zero-sum.  That understanding probably accounts for the ‘me too’ effect in healthcare, as once one system purchases a gamma knife, the other system will to, as they are unwilling to let the competitor capture those lives with the resulting profit strengthening one system over the other.  

But this zero-sum mentality trickles down as well from the CEO level to employees, particularly middle management.  Consider the service line manager – given a fixed budget, bonused on cost savings vs that budget ceiling.  You have value-added services that earn revenue.  However, you also have compliance-related non-value add mandatory services which are essentially costs.  What’s one way to improve the service line budget?  By keeping the valued added work and pawning off the non-value added work as much as possible on someone else.  By having your clinicians bill separately for services, and requiring by medical staff privileges that ‘cherry picking’ is not allowed, you make sure your clinicians will provide services to the indigent as well as the insured.  But you don’t have to pay your clinicians for that work.  By requiring department chairpeople to design standard orders, you avoid having to hire consultants to do the same thing.  Cost-shifting onto the non-employed physician is a well-known phenomenon.  Don’t think that it doesn’t work the other way, however!  On a busy friday afternoon, a family practitioner sends a complicated elderly patient to the ER with a weak complaint which requires evaluation.  When it is time to discharge the patient, the family members can’t be found and the physician, who does not have privileges at the hospital, won’t answer the phone.  An economist would argue that each of these individuals acted in their own best interest, but the cost to the patient and the system, as well as the payor, is high.

As physicians are employed in the hospital system,  the situation gets more complex.  Cost-shifting behavior dies slowly, but the mid-level administrator is merely shifting costs within the system to another service line manager to meet their own budgetary  or productivity goals.  Without an institutional understanding of why this behavior is maladaptive, and management processes in place to make sure this does not happen, the result is that employed physician is cost shifted upon – and that person has lost the ability to cost-shift herself back to maintain equilibrium by virtue of employment.  This is a problem, because it can cause physician dissatisfaction, a declining quality of care, and ultimately physician burnout.  And currently, there does not seem to be any governance model in place to prevent this (At least, I’m unaware of them).  What will ultimately happen is service lines will be missing key players, resulting in missed revenue opportunities for the system – essentially giving their competition the edge – in light of positive budgets and productivity goals.  This will leave most executives scratching their heads, as the relationship is not directly seen.  The bottom line is that you can’t cost shift onto yourself.  Systems employing physicians in significant numbers would be wise to learn this quickly.

Resource Misallocation in Graduate Medical Education


I hope it is clear to anyone reading this blog that this is a terrible waste of human capital.

If it becomes widely known that you need to spend $320,000 on four years of medical school just to compete for a shot at a residency, the ‘best and brightest’ will take one look at that, say “No thank you” and re-orient to careers that do not subject them to an inordinate degree of personal and professional risk.  Medical students will then be picked from 1) the truly wealthy, 2) the uninformed, and 3) the desperate, looking for a lottery ticket.

I am mentoring a young physician who falls into that gap and has been unable to secure an internship.  Once upon a time, this physician would have slid readily into a less competitive specialty – pediatrics, family practice, etc…  But now, their ability to practice medicine in the future is really in jeopardy.  This is a bright person with an ivy-league background and a winning personality, but coming from a lower-tier medical school.  Their dream of being a physician is at risk of becoming a nightmare.  And the terrible thing is that this individual’s story is not a fluke any more.  The terrible state of Graduate Medical Education (GME) in the United States needs to be addressed.

P.S. Any program directors needing to fill a slot with a great intern, contact me.


Productivity in medicine – what’s real and what’s fake?

Let’s think about provider productivity.  As an armchair economist, I apologize to any PhD economists who feel I am oversimplifying things.

Why is productivity good?  It has enabled the standard of living increase over the last 200 years.  Economic output is tied to two variables: the number of individuals producing goods, and how many goods and services they can produce – productivity.  Technology supercharges productivity.   50 member platform companies now outproduce the corporation of 40 years ago which took a small army of people to achieve a lower output.  We live better lives because of productivity.

We strive for productivity in health care.  More patients seen per hour, more patients treated.  Simple enough.  But productivity focused on N(#) of patients seen per hour does not necessarily maintain quality of care as that metric increases.  A study of back office workers in banking validated that when the workers were overloaded, they sped up, but the quality of their work decreased (defects).  Banking is not healthcare, granted, but in finance defects are pretty quickly recognized and corrected [“Excuse me, but where is my money?”].  As to patient outcome, defects may take longer to show up and be more difficult to attribute to any one factor.  Providers usually have a differential diagnosis for their patient’s presenting complaints.   A careful review of the history and medical record can significantly narrow the differential.  Physician extenders can allow providers to see patients more effectively, with routine care shunted to the extender.  However, for a harried clinician, testing can also be used as a physician extender of sorts.  It increases diagnostic accuracy, at a cost to the patient (monetary and time) and the payor (monetary).  It is hardly fraudulent.  However, is it waste?  And since it usually requires a repeat visit, is it rework?  Possibly yes, to both.

The six-minute per encounter clinician who uses testing as a physician extender will likely have higher RVU production than one who diligently reviews the medical record for a half-an-hour and sees only 10 patients a day.  But who is providing better care?  If outcomes are evaluated, I would suspect that there is either no difference between the two or a slight outcome measure favoring the higher testing provider.  An analysis to judge whether the cost/benefit ratio is justified would probably be necessary.  Ultimately, if you account for all costs on the system, the provider that causes more defects, waste, and re-work is usually less efficient on aggregate, even though individually measured productivity may be high.  See: ‘The measure is the metric‘.  Right now, insurers are data mining to see which providers have best outcomes and lowest costs for specific disease processes, and will steer patients preferentially to them (Aetna CEO, keynote speech HIMSS 2014).

One of my real concerns is that we are training an entire generation of providers in this volume-oriented, RVU-production approach.  These folks may be high performers now, but when the value shift comes, these providers are going to have to re-learn a whole new set of skills.  More worrisome, there are entire practices that are being optimized under six sigma processes for greatest productivity.  Such a practice will have a real problem adapting to value-based care, because it represents a cultural shift.  It might affect the ability of a health system to pivot from volume to value, with resulting loss of competitiveness.

In the volume to value world, there are two types of productivity:

  • Fake productivity: High RVU generators who do so by cost shifting, waste, re-work, defects.
  • True productivity: Consistent RVU generators who follow efficient testing, appropriate # of follow-up visits, and have the good outcomes to prove it.

I am sure that most providers want to work in the space of real productivity – after all, it represents the ideal model learned as students.   Fake productivity is simply a maladaptive response to external pressures, and shouldn’t be conflated with True productivity.

Things to think about in transitioning from a private practice to hospital-owned practice – productivity, pay, and metrics

Browsing the other day I came across a question asking about salary metrics.  A multi-specialty group (MSG) was looking at a sale to a hospital.  All providers were on flat salary.  (uh oh)   
Paying everyone a flat salary in a single specialty group (SSG) communicates that you are being paid for your time, not necessarily productivity.  This may work in some models (dept. of health clinics, clinics targeting the underserved, etc…) but in a SSG with multiple physicians, it will eventually cause problems as 1) different physicians have different productivity naturally and 2) paying a flat salary regardless of income is fiscally dangerous – you can’t spend more than you earn.   This approach may flatten productivity in a ‘lowest common denominator effect’.  

At the other extreme, is ‘eat what you kill’.  This creates a vicious practice environment where partners fight over high RVU & highly paid work, and ‘dump’ the unpaid or poorly paid work on each other, other clinicians, anyone they can!  Younger, less connected members of the group are taken advantage of by older, savvier partners.  This kind of practice (and they do exist!) is where medicine gets its reputation of “eating its young.”  It is an anti-collegial system, and results in high turnover, a lower level of overall care, possible legal risk, and ultimately a lawsuit when the providers split up.

So, how to resolve this problem?

1.  Its important that you KNOW your provider’s productivity.  How many RVU’s?  How many patients seen?  How many procedures?  What are their charges?  What are their receivables? You need to measure these items.  Billing records may give a reasonable approximation.  Consider basing productivity on charges, not revenue, as different payor mixes may have different reimbursement, and swapping a provider to another site/shift might account for differences in recovered revenue.  Also see discussion below in #3 for philosophy.

2.   Once you know the average productivity of the providers, then you can establish the level of salary from MGMA for a group of that % of productivity.  Consider establishing the base salary at a slightly lower level (i.e. if average group productivity is 65%tile – 85%tile, set your base salary at the 65th%tile not the 75th%tile mean) so that less productive members of the group are not dismissed at the first opportunity if they are not meeting productivity measures.  In a MSG setting, it might be better to treat it as a bunch of single specialty group contracts negotiated under a master agreement.

3.   Establish a bonus based upon excess RVU’s to encourage productivity.  Be careful here, as solely basing the bonus on RVU’s can cause the group to lose cohesiveness and collegiality.  Even better, if you can model it correctly, use a hybrid model of RVU’s, # of patients seen, total $ amount of charges.   This last part is important, as one of the big advantages of being a hospital-owned group is the ability to be separated (in theory) from accounts receivable.  Bottom line – providers are doing the work, and the hospital is doing the collection.  You (the providers) need to be paid for your work & the hospital needs to collect.  If the hospital is not able to collect, that is beyond your ability to control in a hospital-owned practice, and ultimately not your responsibility (although you must do everything in your power to help them collect by coding properly & compliantly).  It is a shift in thinking from shareholder to employee.  One neat thing that you can do here as a MSG is set a ‘group bonus’ tied to the overall productivity of EVERYONE now in the MSG swept into the hospital group and a separate ‘individual bonus’.  That might go a long way to maintain the culture which existed in the MSG and keep the providers happier. 

4.  Nobody likes to do work that they are not paid for.  So for administrative duties (chairmanships, committees, etc…) negotiate a small(er) bonus for that specific work.

5.  There are quality measures that need to be met under meaningful use criteria, and the hospital leadership may have set their own performance measures.  There should be a small bonus for meeting these measures as well as a small demerit for not meeting them. (+/- 0-2%?)  This should modify the overall group and individual productivity bonus to discourage folks from boosting RVU’s at the expense of quality measures.

6.  For call, you might be wise to negotiate a flat rate per call with the hospital (specialty-specific).  That way, those who hate call can ‘sell’ their call to those who like to take call or who are hungrier for earnings.  If you do so, you MAY need to hold the call earnings out of the RVU pool as otherwise those who take more call will have more RVU’s and skew the bonus pool.  However, the calculation may be difficult to do.  

7.  Finally, once you go through this process you can standardize a day’s pay, and those who want to work less can buy vacation days from those who want to work more.  This is a nice option if available.

8.   Be really clear about the metrics established for performance evaluation, promotion, and bonuses.  Try to make it fair but don’t provide incentives for uncollegial behavior, substandard care, etc…  It will save money and heartache later.  See previous post on “The measure is the metric.”  Solely basing employment on RVU targets is risky.

Z.B.  While I think its fine to ask on the net about options, there is no substitute for specific, expert advice from someone who has gone through this process before – preferably multiple times!  Being that a MSG has the income of multiple physicians, I think that it would be wise for them to hire a consultant who has guided groups through this kind of transition and can evaluate the practice intimately under a NDA and provide specific recommendations (which this post is emphatically NOT).  Perhaps the questions and comments above may serve as a very rough beginning of a process which will lead to a successful cash-out and transition from private practice to hospital-owned practice. 

Where I’m going with this blog

Quick thought for this Monday morning….


I’m taking Trevor Hastie’s and Robert Tibshirani’s fantastic Statistical Learning course online through Stanford.  Not for the faint of heart or small in mind.  (Yep –  THIS is what I do for fun.)
The slide here is great – and shows the danger of complex models and overfitting.   For the data scientists among you, it comes out from cross validation.  If any system traders or finance people are reading, think walk-forward analysis.  

Basically, what the graph says is that when you apply a better and better model with higher levels of refinement to your system, you ‘fit’ your established data more accurately.  However, because the system is more complex, it is more rigid and less flexible (degrees of freedom, anyone?) and less resilient.  Tracks the data better, but works less well in practice.  That’s why the red line starts going up again as the scale of complexity goes from low to high.  Does this resonate with any of the process improvement (PI) people who are six-sigma trained?  Once you scoop up the low-hanging fruit and pass that first or second sigma in iteration, things get tougher.  A lot tougher.  

This is the fallacy of curve-fitting (also known as overfitting).  More in this case is less, as the model fails to be predictive.  

Where I’m going in this blog is to integrate some of these concepts with resource allocation problems in healthcare; and their resulting effect on patient care.   I’m particularly interested in applying these techniques  to predictive analytics in healthcare.  I think we can learn a great deal from the practical applications of these computational statistic tools which have been successfully applied (I know because I started my career doing it) to the markets on Wall St.   What medicine can learn from Wall St. is a whole topic that I intend to cover, probably in a series of posts. But healthcare is not the markets, and can’t be approached entirely similarly.   The costs of error are catastrophic in terms of lives – it’s not just about money.  

I hope you stay with me as I develop this theme. 

Image from :

MIT webinar on systems thinking for patient wait times

This should be interesting tomorrow….

This is just a quick post, but these guys are obviously coming up with similar thoughts to the ones I am having.  I am curious to know if they are merely following a six-sigma “low hanging fruit” approach or have started iterative refinement of processes with #timestamped data, possibly with workflow modification (and adaptation!!!) as well.  It is hard to come across competitive data on what people are doing out there – some aren’t eager to disclose competitive information, vs. not really understanding what is being asked (more likely).

And ultimately I am curious to see how they are modeling things, if at all.  Is it a Generalized Linear Model, or are they doing something different?

Excited to see what these gents have come up with – bright people – one from MIT, the other from Amazon.

Hat tip to Paul Levy’s Blog and @Docweighsin on twitter for alerting me about this.  Paul Levy is a consistent source of great, high-level thinking on the internet.

Dear Doctor, (letter to a Doctor)

Physician reviewing records

This is a post from a person I interact with on social media. It has been heavily modified to keep anonymity. I have obtained express consent from this person to share their views here.

Dear Dr. — Thanks for seeing my child today & conducting a comprehensive exam. We were pleased with your care & the recommendations received.

However, please work with your staff on:
1 -Don’t tell me ‘1 hour’ if I ask how long the the appointment will last and then expect me to be happy after more than three.   Yes – I do know I will have to wait  – a range would be helpful.
2 – When called to reconfirm by your staff, I asked if they had all of our reports sent 2 months ago which were printed for you (it’s a little complicated).   Don’t have them tell me ‘yes’ when the answer was ‘NO’.   Putting a ‘see me’ post it note on the file from a staff member who is out of the office is not helpful.
3 – You are excellent in what you do.  I’m happy to pay for your knowledge and expertise but not your data entry skills (see above).
4 – When I explain to your staff that my child is uncomfortable going to physician’s offices and I need to prepare him about what to expect, please don’t giggle.  Is this the first time your staff has been asked this question?  I can’t believe that.

Thank you.



-A friend once sent a bill to his doctor for making him wait 3 hours.

-I hear you . Waiting forever is the worst! Some health professionals need to brush up on their interpersonal skills.

-(We) were just talking about the medical practitioners we’ve left over the years…because of their staff!!

-…staff was really frustrating.  …tried to give feedback constructively and professionally but the attitude was unreal.


Can anyone not relate to this?  (Unless you are a practicing physician or administrator and you are so busy you have no time to go to the doctor!)  I view this as a systems failure.  The processes to make sure that this patient had an excellent experience were not there – the Doctor seems to being doing all he can to make the experience great (except for the ubiquitous data-entry EMR curse that patients hate as much as physicians!), but the staff undermines his efforts and this visit goes squarely into the negative category.  Regardless of where you want to place accountability (the staff, the physician, the office manager, the administrator), the root cause of this negative experience could be looked at and improved.

What the patient (patient’s parent/responsible party) wanted in this circumstance was:

  1. Accurate scheduling (responsible booking, integration with MD’s calendar)
  2. Accurate information (saying “you should block off your afternoon, but we will try to get you out in an hour” would go a long way here)
  3. No data entry (hire a scribe or switch your EMR system!)
  4. Transmissible Review of information by a staff member (no “see me” post-its – that’s poor continuity of care)
  5. To be treated with respect and dignity (NO giggling or attitude).

The last item is the most concerning – I know that we are starting to recognize ‘compassion fatigue’ and ‘burnout’ in docs in increasing numbers, and it almost certainly crosses over to support staff.  But this offending staff needs to be trained/educated, or shown the door.  Someone else’s discomfort is never a cause for a healthcare staffer’s entertainment.  Better to create systems and processes that rein in the chaos and allow these staffers to feel less besieged and give a level of care that supports the hard-working doctor’s efforts, not negates them.


mHealth – what is the path forward?

mhealthI’m only getting started in learning about the mHealth community and applications.  But from the (limited) amount I can gather about this somewhat distributed ecosystem, it seems to be this:

1.  Lots of mHealth Apps out there.  
2.  Many designed without serious clinical input (i.e. mediocre or wrong information)
3.  Many designed without attention to UI (user interface) or UX (user eXperience).
4.  Apps that require you to plug in EVERYTHING to a database suggest a failure of design.
5.  Few real users.

That’s why we’re not seeing a ton of mHealth apps exploding on the marketplace and multiple IPO’s.  Perhaps I’m being a bit harsh here, as we just might be early to the game.  But where is the end result?

1.  Lip service aside, no one wants to PAY for wellness.  And more concerningly, there is no evidence that wellness works (1)
2.  Poor UI/UX, as described above.  
3.  No integration into patient portals or hospital EHR’s for the majority of users.  
4.  Data from apps not readily available to providers.
5.  No data from application to show that it is useful to clinicians.

Specifically targeting point #1 – aside from the motivated quantified health folks, how do we get the right app into the hands of the right people?  From my point of view, a progressive vertically integrated system – insurer – hospital – provider all rolled up into one – could target (hotspot) an area of cost excess or refractoriness in meeting ratios or a MU requirement.  Then, BUY the app & developers.  INTEGRATE it into the systems EMR in a way that the providers feel is useful and that should give a triple aim result.  Then DISTRIBUTE it to your hotspot in such a way that group members are motivated to use it & TRACK the results.  Win-win for everyone.

At the recent HIMSS 2014 conference, the bulk of the discussion of mHealth was on mobile data management to meet HIPAA compliance.  Sensible, but suggests that the critical mass for actual clinical applications is not there, despite recent FDA liberation of rules for health apps.  

Anyone more connected in the mHealth community want to disagree with me and educate me in the process?

Source: Reuters, RAND corp.


Triple Aim? How about the Quad Win!

I’ve heard a lot about the Triple Aim lately – the goal of:
Improving patient experience through both quality and satisfaction.
Improving population health (care)
Reducing the per capita cost of healthcare

Well, I think we should be thinking more inclusively, and for that I propose the Quadruple Win:
Better Patient Satisfaction
Better Patient Outcomes
Better Provider Satisfaction
Reduced Cost

The difference is in structuring meeting the Triple Aim to include Provider Satisfaction. 

It is beyond the scope of this discussion, but the chatter from physicians is more negative than usual.  Granted, a lot of physicians are ‘glass half-empty’ types, but even so, the pervasive uncertainty about the future of their careers is widespread and palpable.  A number of articles about physician ‘burnout’ have been hitting the rounds lately, even from venerable sources like the New York Times. (1)  How does a profession that prides itself on hard work and a single-minded devotion to their patients go from discussion of exciting new medical cures to 50% burnout rates, even starting in medical school?  I find that difficult to understand – on a personal level, I couldn’t wait to get to work when I was in medical school on the wards – it was just so exciting and interesting!  So how did we get from there to this?

Perhaps mandating the triple aim without including physician well-being (provider satisfaction) has something to do with it?

More on this to come.