Version 1 Care Management to Prevent Hospital Readmission Fails (Unsurprisingly)

Does this high profile randomized study "prove" that telephone follow-up of recently discharged inpatients fails to prevent readmissions? 

Should hospital leaders reconsider care management programs aimed at reducing readmissions?

Hardly, says the Population Health Blog.

Here's how the study was designed:

To be eligible, patients had to be aged 55 or older and without mental illness or serious cancer. They also had to be able to use a telephone. If the patient and their doctor agreed, patients were then randomly assigned to either:

1) "usual" care that consisted of a pre-discharge review of medications, follow up and other instructions plus a 10 day medication supply, or

2) "intervention" care that was comprised of pre-discharge disease-specific education using motivational interviewing, personalized notification of the primary care physician for follow-up, a medication schedule, an in-person follow-up by a registered nurse within 24 hours and follow-up telephone calls on days 1-3 and 6-10 after discharge.

Over 6300 patients were reviewed, 1781 patients were considered and 700 were enrolled in the study. 679 patients completed 30 days, 581 patients completed 90 days and 561 patients completed 180 days of follow-up.  The mean age of the study population was 66 years, 56% had mild cognitive impairment, 33% had visited an emergency room in the prior six months and 62% used English as their primary language. 

In the intervention group, nurses managed to complete their two phone calls 83% of the time.

Results?

"There were no statistically significant differences in the number of ED visits or readmissions between the intervention and usual care groups at 30 days (0.33 vs. 0.26 per person-month; 112 vs. 89 events), 90 days (0.23 vs. 0.20 per person-month; 238 vs. 203 events), or 180 days (0.20 vs. 0.18 per person-month; 392 vs. 370 events)." 

There was also no different in the number of primary care visits between the two groups.

Ouch.

The authors speculate that this patient population already had a high level of support from primary care providers and good access to medications.  In addition, a high prevalence of cognitive impairment may have blunted the nurse interventions.  Last but not least, the authors state that further reductions in ED visits or readmissions may require more in-person home visits in lieu of just telephone calls.

The Population Health Blog offers another thought:  the study was doomed from the start.

Years ago, the Ver. 1.0 "disease management" vendors learned the hard way that aggressively "calling" every patient with did not reduce complications, costs or health care utilization. 

The study described above was a reprise of that long discredited approach. Calling every person being discharged from a hospital may help some patients, but not all. 

Since that time, "population health" vendors have discovered risk stratification. By restricting their in-person and telephonic follow-up to patients discovered to be at greatest risk by advances in"big data" analytics, resources can be better focused on the patients who are most likely to benefit and the likelihood of a return on investment is accordingly increased.

And it's not like this is rocket science.  Surveys and clinical algorithms like this and this respectively can help identify recently discharged patients at high risk of readmission. 

If the study above had incorporated this approach and only enrolled the high risk patients, they might have had a positive study. 

That's the real lesson for hospital leaders and their care management programs.

Image from Wikipedia

Doubling Down on Accountable Care Organizations and Health Information Networks

Want to achieve effective health care, reduced costs, increased quality, population health, widespread prevention and seamless health information access? 

It's easy, says  this article in Population Health Management: mix one part PHO with one part HRB to create a HAPPI.

The Population Health Blog was confused too, but that's what's proposed by three smart academics from Johns Hopkins, Arizona State University and UC Berkeley.

As the PHB understands it, Population Health Organizations (PHOs) would be responsible for all medical, public health, community and social services in a defined geographic area and coordinate them with local education, housing and labor. Much of it would be paid for by a pooled risk-adjusted global or capitated payment (budget) from all insurers.

Each organization would be paired with a Health Record Bank (HRB), which would act as a huge data warehouse that not only stores all medical information, but any other publically available information on every individual enrolled in the PHO. The HRBs would be owned and operated by "trusted custodial organizations." Data access would be ultimately controlled by each patient.

The authors believe that patient payments would be a source of additional revenue for their PHOs. Examples include buying "apps" that are tailored to their individual health needs, or selling their personal health information, especially if it means helping physicians buy an electronic health record or access cutting edge research.

Combine a PHO and HRB and you have a Health and Prevention Promotion Initiative (HAPPI). Its size and scale would warrant contributions from community and provider organizations "without the need for additional reimbursement or outside funding." It would efficiently "align incentives" for insurers, hospitals and ACOs - with money left over for prevention, care coordination, decision support and a learning health system.

Breathtaking, isn't it?  If any PHB readers thought accountable care organizations (ACOs) and health information networks (HINs) weren't big enough, along comes Tyrannosaurus rex-sized PHOs, HRBs and HAPPIs. 

The PHB worries that while we'd want to see how pint-sized ACOs (not a slam dunk) and HINs (likewise not a slam dunk) perform before we apply the massive steroid doses, the opposite could happen: their messy failure could be just the justification for doubling down and going even bigger. 

As pointed out in a recent Wall Street Journal Notable and Quotable:

Economist Michael Munger writing in the Freeman, Aug. 11:

When I am discussing the state with my [academic] colleagues, it's not long before I realize that, for them, almost without exception, the State is a unicorn. I come from the Public Choice tradition, which tends to emphasize consequentialist arguments more than natural rights, and so the distinction is particularly important for me. My friends generally dislike politicians, find democracy messy and distasteful, and object to the brutality and coercive excesses of foreign wars, the war on drugs, and the spying of the NSA.

 

But their solution is, without exception, to expand the power of "the State." That seems literally insane to me—a non sequitur of such monstrous proportions that I had trouble taking it seriously.

 

Then I realized that they want a kind of unicorn, a State that has the properties, motivations, knowledge, and abilities that they can imagine for it. When I finally realized that we were talking past each other, I felt kind of dumb. Because essentially this very realization—that people who favor expansion of government imagine a State different from the one possible in the physical world—has been a core part of the argument made by classical liberals for at least three hundred years.

Image from Wikipedia

Commitment Devices, Behavior Change and Population Health

A new addition to the
behavior change tool box

Whenever the Population Health Blog encountered a tobacco user in its clinic, it would gauge the patient's readiness to quit. For those patients who were ready, it then established a future "quit date" (to facilitate planning), a "contract" (a jointly signed prescription for display on the home fridge) and advice to use any money savings (tobacco is expensive) for a nice reward once seven days of success (for example, a restaurant dinner) was achieved.

The PHB didn't know it at the time, but that seven-day reward was a variation of a "commitment device."  That's what it learned after reading this just-published JAMA manuscript by Todd Rogers and colleagues.

Commitment devices are a way that "present" persons can commit their "future selves" to a sufficient level of needed behavior change.  The threat of a penalty, such as the loss of a night out on the town, imposes a limit on future choices and makes success more likely. 

Other examples of commitment devices described by the authors include applying cash to a success contract (for example, employers could link a bonus to participation in a exercise program that would otherwise be lost), "temptation" bundling that limits access to a gratifying experience in exchange for "consistent" behaviors (used with repeated success by the crafty PHB spouse), limiting bad choices to small packages (smaller portion sizes) and partnering (to avoid disappointing a buddy who shares the commitment).

In retrospect, "commitment devices" have been used in population health for decades.  As Rogers et al point out, however, despite some good research on how effective this approach is, they're generally underused by providers and patients.  One potential way to overcome that is to offer them routinely on an "opt-out" basis, 401k savings-plan style.  The authors also point out that a series of commitment devices on a longitudinal basis could be used to blunt drop outs and maintain long-term behavior change. Last but not least, leveraging social networks with or without handheld "apps" remains an area ripe for future research.

As medical homes spread and shared-risk payment reforms gain traction, the art and science of commitment devices will likely grow. Not only is it a cool piece of insider jargon ("Hey, Mary, I like this care management proposal, but have you any plans to develop commitment devices?"), but any addition to the behavior-change tool box can only help.

Image from Wikipedia

I'm From CMS and I'm Here to Help

Writing in JAMA "online first," CMS Administrator Tavenner and colleagues offer a payment reform "framework" that includes "multipayer collaboration."  The article is wonky, so the Population Health Blog dons its universal adminispeak translator so us normal humans can better understand what CMS is up to.

According to the writers, CMS has a history of innovatively implementing reforms that were later adapted by other insurers. The most famous example is the hospital "DRG" system that, starting in 1983, paid for a diagnosis in lieu of a daily room rate.  Suddenly, hospitals had an incentive to shorten hospital stays, which is precisely what happened in the years that followed.

Buoyed by this success, the authors describe the merits of championing Medicare's transition from "category 1" fee-for-service without any link to quality to "category 4" population-based payments that are linked to quality. And, as CMS embarks on this excellent payment journey toward accountable care, they'll get other commercial insurers to mirror their efforts by:

"Being conveners" as in "working with" other insurers in a region or a state to implement large payment reforms.  Working with may include grants;

"Incentivizing," as in requiring the participation of other payers prior to funding any large pilot programs.

"Working with states" to implement additional reforms, when the state has sufficient influence over the health insurance or delivery system.

The Population Health Blog's take:

"Category 4 population-based payments" are a form of capitation that are ultimately designed to transfer insurance risk from CMS to providers. The PHB hopes the bureaucrats at CMS are aware of the risk re-introducing some 1990s-style managed care abuses. 
 
What also goes unmentioned by the JAMA article are examples of CMS payment reform unintentionally gone awry, including RVUs, regional payment variation and the SGR with lingering fraud. While CMS has had its successes, it's also had more than its share of problems.  Time will tell which track record will apply to population-based payments.

Convening was an art developed by Medicaid programs.

Ms. Tavenner implies that population-based payments (a form of capitation) are intrinsically linked to quality.  Nothing could be further from the truth, since it's possible to reward quality while also relying on a FFS methodology. 

Accountable population-based care remains a large experiment.  Ms. Tavenner implies that there is an aura of inevitability.  The PHB learned long ago that the sign of a good plan is an exit strategy in case things go south.  The PHB didn't read that here.