Archives for

Mandatory Infrastructure for a Health System ACO (Spoiler Alert: It is NOT Data Analytics)

I am more than a little surprised by the lack of attention given to the fundamentals of actuarial analysis and complex care management in discussions of ACO management.  I see repetitive references to the criticality of deep data analytics, and to implement “data driven” decision making.  But let me make a couple of somewhat obvious observations:

  • It is likely true that the only way to make long-term real margin in the business of population health management, the provider organization needs to be at full risk for the management of patient costs (this is why health plans predictably make money). A particular physician group can certainly nibble around the edges of cost by managing referrals, attempting to influence ED utilization and being careful with prescribing, but this does not mitigate the risk of an outlier.
  • If a provider group is actually going to take on full risk (not merely “risk,” but full risk) for a patient population, the population would need to be actuarially sound. The average MSSP ACO is less than 20,000 patients.  MSSP patients are not even really “members” yet (since the patients may not know they are in an ACO), but that is a future possibility.  Nevertheless, even if they were members, 20,000 patients are not remotely actuarially sound.  “Sound” in this context means that the total costs of a population are credibly projectible into the future, within an acceptable margin.
  • The top 1% of patients average costs of about $200,000 per year. If a group of 20 primary care physicians are managing 20,000 patients get and “extra” three of those top 1% patients, that is an incremental $600,000 reduction in their base pay.  That looks like $30,000 per physician to me.  If they get an extra 6 of the top 1% (again, out of 20,000 patients) they lose $60,000 each.  That would be about 20% of the average primary care physician’s compensation. Ouch.
  • I guess those physicians are not buying expensive groceries that year.
  • …Unless the group buys reinsurance (more about that below).


  • It is NOT usually difficult for physicians to identify the patients who are at risk for becoming outliers. Most primary care physicians can identify their top 10 risky patients off of the top of their heads.  The question is NOT who is at risk.  The question is what to DO about the risk.
  • Most physician organizations are not large enough to deploy care management resources (e.g., complex case managers, in-home clinicians, clinical call centers, etc.) to mitigate the risks of their complex patients.
  • They do not need more information about their patients. They need more resources to care for the outliers.  It is not that deeper analytics are not useful.  The issue is that deep data analytics are not the largest need.


The dearth of discussion about reinsurance in the context of managing risk in patient populations is more than a little surprising.  Large reinsurers manage the risk that cannot be handled by parties at risk for a small population.  Many small health plans (100,000 to 300,000 members) buy reinsurance to mitigate the risk of an unexpected debacle in any one year.  Reinsurers aggregate risk for a number of small populations and manage the risk of a larger population.  The size of the reinsurance premium is based on the degree of risk passed off to the reinsurer.  There are several models for the reinsurance framework, but for now that is below the noise.

The problem for providers (and reinsurers) is that a very small number of patients consume most of the costs.  The top 2% of patients are about 32% of all costs in the US.  The average cost of a patient in the tops 1% is about $200,000.  The next percentile down is about $120,000.  These two brackets are the majority of manageable costs in the population.  If a physician group buys reinsurance for patients that cost more than $100,000 per patient (for example) they will likely have successfully mitigated the risk of being unable to afford groceries the following year, but they have also obliterated most opportunities for cost savings.  Further, a physician group, unless very large, would not likely have resources to deploy non-physician clinicians in such a way to manage the risks of these high-risk patients.  What are physicians to do?

Health Systems Could Step Into This Breach

Let’s assume (for the sake of argument) that the minimum patient population that be actuarially sound is 200,000 members.  (I do understand that commercial populations could be smaller, and Medicare populations could be bigger, but let’s finesse that for the moment). “Actuarially sound” in this context means:

  • The costs are credibly predictable within an “insurable” band, and
  • The acquisition of reinsurance can be set at such a point that the attachment points for reinsurance do not obliterate any opportunities for savings.

What is the structure that a health system could deploy to assist primary care physicians with their risk-bearing populations?  There are four key steps in the process:

  • Aggregate risk-bearing populations- The health system should act as the intermediary reinsurer (yes, this requires an insurance license and some new skills). Health systems can acquire a reinsurance license and charge risk-bearing groups insurance fees for their respective populations.  The health system could offer internal reinsurance to ten or more 20,000-member risk-bearing populations.  The health system would then buy reinsurance at a much higher attachment point (think $300,000) than could be afforded by the primary care groups, since the health system is now managing a 200,000+ member risk-bearing population.  The health system would then use the reinsurance fees from the various risk-bearing primary care groups to subsidize key centralized care management resources.  The health system in now in a position to directly contract for incremental risk-bearing populations as well.
  • Deploy centralized care management resources- The health system should deploy common resources to support the members that are explicitly in risk-bearing populations. The centralized resources would likely include:
    1. A clinical call center– The call center would be staffed by clinicians with access to patient EMRs. The primary objective is to be an immediate care resource for high-risk patients.
    2. A complex case management team– This team would have senior care management resources (typically senior care management nursing staff) managing a pyramid of less senior clinical staff (social workers, home aides, etc.) and specialty clinical resources (e.g., clinical pharmacists) all under the direction of a senior medical officer.
    3. Services to mitigate ED over-utilization– This is highly geography specific, but might include 24 hour primary care offices to handle non-emergent overuse of ED resources by risk-bearing population members.
  • Train the physician groups to integrate care with the centralized resources- This includes not only definition of care coordination processes between primary care and the centralized care management resources. It also includes aligning incentives such that the groups that are effective in mitigating care costs would experience a reduction in internal reinsurance costs the following year.
  • Improve, improve, improve- Any of you that have attempted any of this know that this is actually really difficult. But this is likely the only viable path to effective management of care costs for the majority of US primary care physicians.  The rare physician groups that are at risk for over 200,000 to 300,000 patients have some other options, but we can take those as a special case for the moment.

The objective in this effort is for all parties to work together to get the costs of reinsurance down (both for the individual physician groups and for the health system).  Herein likes the majority of margin in the model.  Once a health system is effective in complex care management, the health system can transition to a greater fraction of total patient volume at full risk.  Full risk directly translates into a higher per-member-per-month reimbursement rate from the payers.


My suggestion is that deep, complex data analytics are useful, but fundamentally secondary to deploying the centralized care resources necessary to mitigate the risks of outliers within patient populations.

I look forward to your comments (and, hopefully, arguments).

Ideas for Atul Gawande and the Amazon/Berkshire/JP Morgan Venture

A Framework for Reducing Costs Through Consumer-Centric Care Management

I have given some thought to what I would suggest to Atul Gawande if he were to ask me how he could use the Amazon/Berkshire/JP Morgan (ABJ) joint venture to move the US health care market toward a more cost-controlled model.  I have outlined below some of the specifics that the ABJ venture could implement that might substantially move the market toward a more sustainable cost model. I do think there is a path here.  I think the core objective is consumer empowerment in price decisions.  But, suffice it to say, this would be a big change.

Does consumerism work in health care?

The answer appears to be “yes.”  If we look at care services that are often not covered by insurance, (e.g., infertility treatment, most cosmetic surgery, lasik surgery and a substantial amount of dental work), we find a couple of common characteristics:

  • The prices are fixed and known in advance,
  • The services have warranties,
  • The services have quality metrics established by the provider (as opposed to the government), and, most importantly,
  • The prices for the services have decreased in real terms over time.

The markets aren’t perfect by any means.  It can be difficult to find prices for these services because the markets are not particularly orderly.  And warranties and quality metrics are inconsistent.  But prices for these services have gone down over time.  I don’t think there is any example of a service covered by insurance that has gone down in real terms over time.  There are some examples of cost-saving substitutions (e.g., inserting a stent versus doing open heart surgery), but I am unaware of any insured service costs actually falling in real terms.

How could we make this work more broadly, particularly in an employer-based insurance model like the Amazon/Berkshire/JP Morgan (ABJ) venture?

It is a credible question whether we could revise the employer-based indemnification and payment model for care services to better replicate the efficacy of price signals for services that are not insured.

The key point here is that competition among insurers is NOT the problem.  Insurers make about 3% margin on premiums.  If they made 0%, we would still have prices increasing faster than inflation and no one would notice the one-time 3% savings. The objective is to generate provider-to-provider competition, as with the uninsured services above.  I think the possibility is pretty high that we could achieve this.

What would the model look like?

Suppose we deployed an employee coverage model along the following framework:

  • Establish consumer-understandable prices for a substantial fraction of services– Restore consumer-understandable price signals by establishing an on-line market for provider services. The service set would include:
    • Fixed-priced bundles for primary care and medical specialty care (probably as annual contracts). Annual contracts for primary care or medical specialty care is be pretty straightforward.  Some primary care physicians offer annual contract for care now (typically about $900) but they are hard to find because there is no orderly market to find them.
    • Bundles for many common procedures. Some bundles are quite challenging, but many are readily feasible.  Complex procedures might require a bid/proposal framework.
    • Prices for ancillary services (labs, diagnostic imaging, etc.).
    • An on-line marketplace. Services would be offered on the web (presumably via Amazon) to establish an open competitive market for provider services, indexed by geography and any other consumer-centric characteristic (e.g., language preference). Rules for the posting will make consumer comparisons easier. I will refer to this service as “Amazon Health.”


  • Restore consumer price exposure– One of the significant roadblocks to consumer centric market change is the structure of insurance. If consumers have no cost exposure above (for example) $5,000, they will not care about the price of the large number of services above that level.  Further, providers will be unincentivized to compete against other providers, since they will not gain significant benefit for doing so. Their only task is to contract with insurers.  The solution is to redesign the deductible/benefit framework to expose health consumers to costs up to about $50,000.  This level would include approximately 95% of all consumer care events.  Specifically:
    • Overall consumer dollar exposure may not change much (e.g., a benefit design with $0 deductible but $20% copay up to $50,000 would still limit overall out-of-pocket exposure to $10,000). I actually prefer higher deductibles, but that is a separate discussion.
    • The framework should be extensible to other self-insured employers in the same markets to amplify the market of the ABJ venture employees.
    • Employees/members with 20% exposure would likely shop for services based (at least partially) on price.
    • If the initiative catches on in geographies where ABJ is employee-heavy, it could be replicated in other markets if a small number of large employers replicate the ABJ benefit design and use the “Amazon Health” service to aggregate provider care offerings in their geography.


  • Disintermediate the health plans provider networks- Require that all participating payers (that is, those payers that provide administrative services to ABJ venture employees) accept any of the bundled services defined in item 1) above as “in-network” (that is, expenses apply to the deductible and are covered above the deductible) whether or not the provider is formally in the network of the participating plan.  This means that any provider in any geography where ABJ has employees could post a price for a service and be “in network.”


  • Indemnify consumers without using insurance– Establish a consumer payment framework for provider services that limits consumer exposure but does not obliterate the price signals:
    • Fund HSAs for employees
    • For employees that incur costs beyond their HSA totals, provide a loan framework to cover the shortfall.
    • Payments on the loan are limited to a fraction of salary (analogous to the ACA limits on the cost of insurance). Keep in mind that most employees in this model would actually repay the loan.  It is also probable (with the correct benefit design) that the combination of care costs and insurance costs would be lower than the current cost of insurance alone for the vast majority of consumers.

The net result of these four legs is that price signals would be restored, providers would have a marketplace in which to post prices and compete, and the market should see downward pressure on services prices.

There are indeed complexities to be addressed (e.g., likely state waivers for benefit design outside of the ACA, potentially federal waivers for HSA- qualified expenditures, management of employee departures when debt is outstanding, etc.), but the potential is there.



A model to replace the commercial insurance portion of the ACA

In my view, any replacement of the ACA needs to separate the discussion into the three major market segments:  Medicare, Medicaid and commercial.  Commercial in this context includes not just employer based insurance, but also the individual market(s) that were mostly addressed by the ACA through offerings on the several exchanges.

This post will only address the latter, the commercial segment.

In my view, the key problem with runaway costs in the commercial market is that the market has lost most of the feedbacks related to price signals.  The price signals are not related to insurance; they are related to care delivery.  More price competition among insurers will have little effect (since insurers only make perhaps 3% on premiums).  But innovations on the pricing of care delivery could have a material and beneficial effect on costs.

If we look at the rare subsegments of the commercial market where care is not usually covered by insurance (e.g., infertility, lasik surgery, cosmetic surgery, even dentistry) we see that they market behaves “normally.”  “Normal” in this context means that the products (in this case, service bundles) have prices known in advance, the providers compete on quality metrics that are not established by the government, the products have warranties, and the costs for the products have decreased over the last 20 years, even though quality has improved.

I don’t know of an insured service that meets the above criteria for “normal.”

If we were to devise a plan to restore price signals to the provider marketplace, we would first need to remove insurance as the intermediary for most purchases.  How might we accomplish that?  The easy answer is to raise deductibles, but that clearly answers only a fragment of the problem.  But suppose we implement the following model as a framework for replacement of the commercial segment of the ACA:

  1. Keep all existing ACA plans as valid (although it might make sense to relax the limitations on variances by age group)
  2. Allow sales of very high deductible health plans, VHDHPs (think $20,000 deductible).  These would be very inexpensive (likely $1,000 to $2,000 per individual per year).
  3. Request or require plans to offer riders to VHDHPs that cover:
    • Primary care
    • Pharmaceuticals
  4. Allow consumers to contribute to HSAs irrespective of the nature of the health plans they select
  5. Allow consumers with any health plan to borrow money from the federal government at the fed funds rate in any instance where they do not have enough cash (or HSA balance) to cover their deductible for health services.
  6. Limit the repayment of debt to the government to the same cash limit as the ACA (9.5% of income).

If payers offer a primary care bundle as a rider to a VHDHP, it would probably cost between $600 and $1200 depending on age bracket.   The moment that insurers offer such bundles, there would be a cascade of primary care providers offering a service bundle for a lower cost (because they don’t incur claims adjudication costs, and now would have lower billing costs).  Consumers would frequently drop their insurer bundle and buy care directly from their primary care provider.

Most individuals in the country (over 50%) would not incur any care costs other than pharmaceuticals beyond their VHDHP and their primary care bundle (which may or may not have a co-pay).  If pharmaceutical costs are high, they could buy that rider from the insurer (or from a PBM since they would likely jump in to bypass insurers).  That set of products (insurance + care bundle) would cost about half of the existing ACA plan.  Folks would abandon ACA plans once they figure this out.  And that would be fast.

But what about the 20-30% of people who don’t use primary care but use mostly specialty care?  Diabetics seeing endocrinologists?  Ischemic heart disease patients seeing medical cardiologists?

Many of you recognize that physicians talk to each other.  If primary care physicians are billing patients $600 per year for a primary care bundle and NOT incurring insurance billing overhead, it is entirely likely that medical specialists would do the same thing for their patients.  This means that primary care physicians would be competing with primary care physicians on the basis of cost and quality.  Specialists would be competing with specialists on the basis of cost and quality.  Do you suppose a multi-specialty group would offer a bundle for all medical physician services?    Hmmmm.

It would take longer for the facility-based care market to move toward consumer-understandable products.  In the interim, acute services that surpass the $20,000 deductible are covered.  Costs below $20,000 would be addressed by a combination of HSA balances and by borrowing from the federal government.  Incidentally, most of that borrowed money would actually get paid back.

In this framework, no one is worse off than they were with the ACA.  Most patients would see their total costs cut about in half.  And now we have actual provider-centric price competition.  Imagine.

This is a path to move the care delivery market toward true competitive price signals.

I welcome your comments.

Framework for Discussion of Population Health Management IT Tools

I am pleased by the frenzy of investment in tools that support the various elements of accountable care/population health.  I am a little surprised how rarely the advertised tools are detailed by the  business functions that the tools support.  I would suggest there are three broad categories of tools in support of accountable care:

  1. Provider performance measurement
  2. Predictive analytics
  3. Care coordination

Each of these categories has one or more key subcategories.  I think it would be useful to detail the subcategories to assist in IT tool selection.

Provider Performance Measurement Tools

This is the category that has been around the longest, and hence is the most mature.  Typically, these tools aggregate claims and group the claims by care episode.  Grouping of claims by episode gives an organization two fundamental capabilities:

  • If the organization is going to contract for care, the tools give information on how to craft the contract, or indeed whether to contract at all
  • If the tools are used during referral of an active patient, the tools can direct a clinician to select a referral target based on previous cost and quality performance

There are literally dozens of tools in this category.  The problem for most provider organizations is gaining access to a significant fraction of claims history to run the tools.  This is somewhat easier for a health plan (because they have the claims for their members) but a provider organization might need claims from multiple payers about many other providers to make this tool perform successfully.

Predictive analytics

Predictive analytics tools are used for two broad purposes as well:

  • Prospective assessment of costs for a defined population to support actuarial assessment/pricing
  • Identification of individuals likely to be utilizers to target treatment to mitigate the risk of adverse cost or quality events

The claims-based predictive analytics tools are quite good at supporting the first bullet above.  Unfortunately, claims-based analytics tools are poor at the second.  The two problems with claims-based tools are 1) the predictive value is low for individual patients (R-squared is often under 50%) and 2) the claims arrive too late to be of any use (60 days post discharge does not give a lot of insight).

Non-claims-based predictive tools are much more reliable if the objective is to direct active patient care.  Non-claims-based tools would include:

  • Tools to assess probability of readmission of a patient at discharge: This assessment is based on current clinical data (from the chart) and knowledge of the patient’s social circumstances (from interview or home assessment).  This data, although harder to accrue and generally available only to the provider directly, is vastly more useful for predicting utilization.
  • Tools to stratify populations based on global risk:  Nearly any primary care physician can identify the top 3-4% of risky patients in his/her practice.  Risk, in this context is a combination of overall disease burden and weakness of social support mechanisms (primarily family) for the patient.  Patients with high disease burden could be targeted for a deeper home assessment to identify preventable risks or potential family support mechanisms to mitigate the risk of utilization.

It is worth noting that disease burden alone is often not a particularly strong indicator of future utilization, hence inclusion of data about a patient’s home environment is critical.  The non-claims-based tools are often much more manual, but far more effective at identifying the high-utilization patients in advance.

There are many tools to assist in the prediction of population costs, and they work well.  There are very few tools to assist a provider organization in identification of high-risk patients.

Care Coordination Tools

As with the other tools, there are two very different objectives in care coordination:

  • Disease management tools:  These tools establish clinical objectives by disease state for each patient to assist in filling in gaps in care and to support reporting of compliance standards to external entities (payers, CMS).  Disease management tools support compliance reporting and quality management.  They do not drive cost reduction (indeed, they often raise utilization).
  • Case management tools:  These tools focus on specific patients in high-risk categories to manage the risk of excessive utilization or risk of treatment failure in complex conditions.

There are many tools that support disease management functions.   These may be a component of the EMR specific to the clinical setting, or an add-on application to the EMR.  In contrast, case management tools are somewhat more rare in the marketplace.  The problem, as often as not, is that the critical issues in case management are often inter-clinical-setting.  The implementation problem may be large enough that the provider organization is not well situated to “connect” to all of the key clinical settings to enable effective case management.

A Note on HIEs

Some might consider HIEs as an element in care coordination.  In my view, HIEs alone do not “count” as care coordination tools.  They may (or may not) be supportive, but if a tool cannot instantiate an intervention and report on completion of an intervention, it is tough to label it as a care coordination tool.  Most HIEs would fail in this regard.


Compliance is Not Quality

I am a little surprised (and somewhat dismayed) by the perspective of many of the competent clinicians that I speak with about quality management.  Many will speak of the various federal reporting initiatives (e.g., PQRS for Medicare FFS, STARs for Medicare Advantage, CQM for Meaningful Use, etc) and suggest that their quality standards are performing well.

I usually respond with “What quality standards?”

Federally established performance metrics are compliance standards, not quality standards.  Some of them may actually impact quality, but many do not.  Does anyone think that BMW sets their performance standards based on the federal government?  WalMart?  Cisco?  Verizon?  Does BMW compete on their federal mileage rating?  Does WalMart brag about SOX compliance? Does Cisco advertise their unqualified audit?

Quality should generally be something that is understood by the consumer of the product as a value.  I recognize that it is often true that the payer for the product is not the consumer, but the selection of the product is usually a consumer issue.

Were I in an orthopedic group, it would be reasonable for my group to set standards for performance that would be understood by our customers- the patients selecting our group.  We might select major joint device failure rate over 10 years, or pain and range of motion six months post discharge or incidence of  infection.  All of these are items that a consumer of major joint services would understand and value.  We certainly would not expect a consumer to care about his orthopedic surgeon’s HEDIS scores.

Can we stop characterizing compliance as quality and focus on real consumer-understandable quality metrics?  I am looking forward to the development of real, competitive quality metrics.  It will be relatively easy to establish meaningful quality metrics for procedural specialties (although it will be somewhat more difficult to actually collect the relevant data).  It will be slightly more challenging to establish and collect meaningful quality metrics for medical specialties/subspecialties, but it can be done.

Has anyone seen innovative competitive quality metrics to report?


How ought we modify the ACA to improve it?

I think it would be instructive to launch a discussion about how best to repair the ACA (aka Obamacare).  I suggest the following would be useful changes that would be broadly accepted:

1) Permanently remove the employer mandate:  I think most economists think this is bad policy, and probably has a significant influence on restricting job growth and increasing the incentives to shift to part-time workers

2) Permanently remove the individual mandate:  The notion that the individual mandate is ‘required’ is, I think overblown.  If the law were changed to implement eligibility rules that require advance purchase (e.g., must buy insurance by 12/1 of the prior year to get coverage, or be continuously insured the change insurers), that would mitigate the tendency of folks not buying insurance until they are sick

3) Keep the exchanges and community rating, BUT:  relax the 3:1 rate variance (old to young ratio) and allow true rating by age and geography.  Also, erase the requirements that minimum insurance is 60% actuarial value.  Let insurers offer as high a deductible as they would like, as long as it is still community rated.

4) Repeal the IPAB:  I don’t think this will have any discernible impact on cost, and is likely to cause more administrative burden in an already overburdened system

5) Remove the limitations on admin costs (i.e., the  mandatory Medical Loss Ratios):  I don’t think anyone believes this reduces premiums; it just decreases the admin functions that may improve utilization control.  Let the payers compete on premiums.

Let me know what y’all think.


Another View of Patient Stratification Data

I had a chat yesterday with a colleague of mine in Portland.  He had just glanced at some of the pages on this site, and referenced a slide from a presentation by Ernst Berndt, and reported in Paul Levy’s blog on 2/20/2013.  You can see that the overall concentration of spend reported here is similar to my text in the “stratify patients” section.

You can see that the graph shows the top 1% consume about 20.2% of care costs, and the top 5% consume about 47.5%.  I think (but I am not sure) this is the overall US healthcare population (it sounds close).

Thanks for Ernst Berndt and Paul Levy for broadcasting.


Which populations are appropriate for care/case management focus?

Most commonly, patients are identified through predictive risk analytics as potential high utilizers.  These patients are then the focus for case management.  Question for discussion:

  • We would probably expect to use different case management styles (and even tools) for the very high utilizers (e.g, over $50,000 per year) than for the moderate utilizers (e.g., $10,000 to $50,000 per year).  How many “levels” of case management ought we implement?
  • Would we expect to have significantly different interventions between the levels of case management?

Looking forward to comments.

Costs of IT in support of accountable care

We have recently endured another round of reports noting that investment in healthcare IT does not seem to have very significant ROI.  We have had redundant reports noting that:

  • EMRs may not have significant ROI
  • HIEs may or may not be “sustainable”
  • Industry IT costs are still trending upward, even though healthcare IT investment is generally lower than other information intensive industries

I continue to wonder why buyers are surprised.  ROI is fundamentally based on changes in revenue or cost after an implementation.  If clinical tool sets do not have much effect on either, why would we expect an ROI?

Perhaps more humorously, in a fee-for-service environment (in which most physicians are still practice), ROI is more commonly driven by better billing characteristics (e.g., better coding on bills, more accurate ‘upcoding’) than by process improvement.  In fact, process costs tend to rise in ambulatory care when clinical systems are implemented.

Does anyone think that these reports have any bearing at all on the value of clinical tools when deployed in an accountable care environment?  That is, if we are actually economically incentivized to improve the cost/quality profile, wouldn’t the investment in care coordination tools have much more obvious value?



InterCase, LLC
80 NW 99th Ave Portland, OR 97229 US
Phone: 5032960129 Website: