Wednesday, April 20, 2016

Medicare's Comprehensive Primary Care Initiative - A Two Year Report

After all the buzz (for example) around the coming launch of CMS' Comprehensive Primary Care "Plus" program,  the New England Journal of Medicine (or NEJM) just published a "special article" on the original Comprehensive Primary Care (CPC) initiative.

This is important if you think CMS' approach to supporting primary care is the fix for what ails the U.S. health care system.

Population Health Blog readers may recall that two years ago, CMS launched CPC. This is a still ongoing four-year multi-payer study to determine whether primary care that is "turbocharged" with medical home-style capabilities (see here, here and here - see page 8) would increase quality and lower health care costs. 

The term "multi-payer" is important, because CMS recognized that clinics struggled with providing medical home care to some, but not all, patients on the basis of their insurance.  Better to have one standard of care to all patients.

The NEJM article is an analysis of CPC's results after two years. 

To summarize how CPC was set up, 502 clinics (from 978 applicants) across 8 states participated along with a total of 39 other insurers.  In addition to the usual fee schedules, the Medicare and the other insurers paid a per patient severity-based "care management fee" that, on average, ranged from $8 to $40 per beneficiary per month (PBPM). Practices were also promised an additional bonus if, after two years, they reduced health care costs (i.e., shared savings) and improved various quality measures and performed well in surveys about the patients' experience of care.

These CPC practices' outcomes were compared to a propensity matched group of non-participating practices with a similar electronic health record (EHR) infrastructure that cared for a set of patients with similar levels of disease and baseline costs. 30% of these practices had applied but were not accepted in the initiative. The total number of comparison practices was 908.

Results?  Not good.

Aft the end of two years, there was no statistically (p > .05) significant difference in the growth of health care costs between the CPC and control sites.  This was true whether just claims costs were examined (a negligible difference of $11 per patient per month favoring the CPC sites), or whether claims costs plus the additional fees were examined (a difference of $7 favoring the comparison sites).

When patient costs were examined by the burden of disease, there was no indication that more costly patients achieved any savings. 

CPC sites had a statistically significant reduction in outpatient office visits, but not in hospitalizations.

While the difference in claims expense failed to be statistically significant, the total additional fees collected by the participating sites amounted to a financially significant $389,000. This represented a 15% increase in their income

Was quality of care improved?

Patients with diabetes and a high burden of illness were more 3% more (p<.05) likely to receive the recommended follow-up measures to manage their disease. Otherwise, "the initiative did not have significant effects on the processes used as measures of the quality of care for the full sample."

Patient experience of care?

While surveys showed small increases in patient support, "there were no significant effects on other composite measures: ability of patients to obtain timely appointments, care, and information; how well providers communicate with patients; provider’s knowledge of care patient received from other providers; and overall rating of providers by patients."

Yikes. Ouch. Egads.

The authors correctly point out that CPC is a four year program and that it still may be too early to see the impact of the medical home turbocharging.  That was pointed out in the negative one year evaluation.  Maybe something will turn up at three or four years.

In addition, CMS has a lot of other value-based initiatives underway, which may have biased the results.  There may be a "ceiling effect" among the participating sites as well as the control sites, which were already working to reduce (for example) rehospitalizations or pursue the fee schedule modifiers.

It's also important to note that the impact on the other insurers' costs and patient quality was not reported.  It's possible that they saw a benefit.

The PHB's take?

1.  Many care management programs achieve claims reduction with savings (for example) within one to two years.  If CPC hasn't succeeded by now, it probably won't.  And if the just-announced CPC Plus is modelled after this, it's hard to see how that program will turn out any differently.

2. It is possible that, within all the statistical noise, there were some primary care sites with particularly robust approaches to care that did bend the cost curve.  CMS should seek these sites out and find out more about their secret sauce.  More on that in a future post.

2.  If CPC's approach to care is ultimately shown to not bend the curve, what's the problem? 

The PHB continues to believe that one size doesn't fit all and not all patients benefit from care management. Many patients, even those with chronic conditions are quite stable and need minimum attention; some patients are so sick that no intervention will keep them out of emergency rooms and hospitals. As pointed out here, as more and more patients are enrolled in care management, the return on investment can paradoxically go down. Better to focus on patients who are not only at risk, but have "impactable" condition profiles.

In addition, CPC is based on a 5 year-old model of care. Things have changed since then: modern population health brings many more resources to the table.  That not only includes in-depth analytics support (for example, to define those patients who are at greatest risk) but mHealth. For example, there is one innovative company (the PHB's Shameless Commerce Dept. over on the right side of your screen) that provides recently discharged patients with an app-enabled handheld configured to provide close follow-up.  And so on.

3. It may be that care management works best in a managed care setting.  CPC is a study of classic fee-fore-service Medicare beneficiaries with access to any participating Medicare provider. In Medicare managed care, the insurers and their providers have an even larger incentive to maximize quality and lower cost.  If that's the case, CMS - despite their commitment to innovation - may want to get out of the care management business, because they just don't know how to do it.

1 comment:

Michael Cousins said...

Jaan -- Kudos on another great post. And thank you for mentioning impactability. In my view, not enough practitioners/policymakers focus on patients/members who are impactable -- and of course that means that not enough innovation efforts are directed to figuring out how to convert the non-impactable to impactable.

To make matters worse, many vendors who claim to offer predictive models that identify impactable patients are long on marketing and short on science and results.

By way of background for those who may be interested, Iver Juster MD, and I defined the term when we served together on the DMAA's Predictive Modeling Committee. It was published in the DMAA's Dictionary in 2004 and has since caught on. The spelling of the word has changed over the years (from -ible, which was based on the word "susceptible", to -able) but the need for analytics and clinical programs to focus on impactability hasn't changed one bit.