The Lament About the Healthcare "Return on Investment"

The Population Health Blog had time to go back and review this New England Journal article on "return on investment" in healthcare

It's abbreviated "ROI."

In it, academic researchers David Asch, Mark Pauly and Ralph Muller lament that interest in getting a "return" from reducing healthcare utilization is unfair. While it is a sought-after metric in chronic conditions (for e.g., diabetes) it's practically unheard of other care settings (for e.g., cancer care).

The authors point out that may be because:

1) Care of conditions like cancer is very remunerative to providers, so there is little interest in reducing income. In contrast, diabetes has little "top-line" potential;

2) Unlike conditions like diabetes, reimbursement around the "episode of care" following a new diagnosis of cancer explicitly supports a known - and profitable - suite of hospital-clinic services;

3) "Savings" from reduced health care utilization can be complicated by the "back-filling" of empty appointment slots and unfilled beds with other patients with other needs and other sources of income.

There are two solutions.

The first is at the front-end by decreasing (with or without bundling) the reimbursement. That would presumably force the providers to gain care efficiencies that exceed the accompanying lower payments.

The second is at the back-end with "shared savings." This financially rewards providers who can muster efficient episodes of care. In other words, the check is the "ROI."

All good points, but written from the provider perspective.  From the perspective of buyers (businesses and individuals who buy/pay taxes for commercial or government insurance) it's more simple: services flex up to meet generous fee schedules and flex down when payment shrinks.

The right balance between the money and care can be determined by brutal and efficient markets or by all wise and mistake-prone policymakers.  Take your pick, implies the authors, but if it's the latter, the results are preordained.

The PHB would offer three other points on ROI:

1) We've seen this movie before: Using financial incentives to drive fewer hospitalizations, drugs and specialists is perilously close to rewarding the withholding of needed care.

2) Measuring non-events is hard: "ROI" in most healthcare settings is not a classic ratio of income to investment, but savings to investment to savings. The latter is ultimately based on a statistical measure of what doesn't happen vs. the baseline utilization of a large population. It's not easy to discern the "signal" of fewer pricey hospitalizations, fewer expensive drugs, or less need to see costly specialist physicians from the "noise" of healthcare inflation.

3) High health status ≠ low cost: Increasing quality of life is often a function of increased access to costly health care that is often a function of socioeconomic status.  In other words, you get what you pay for in both healthcare and lifestyle.

Which is the PHB's lament It's not a function of "saving" money, but using it wisely.  It's not a matter of ROI, but creating patient-centric value.

Image from Wikipedia

Is $1 Billion a Good Investment for Disease Management? We May Finally Have an Answer

Take two of these and call for
savings in the morning

Anyone familiar with the history of disease management industry will almost certainly can recall Soeren Mattke's 2007 article that provocatively asked Evidence for the Effect of Disease Management: Is $1 Billion a Year a Good Investment? Based on what was known in the peer reviewed literature at that time, Dr. Mattke's answer was a desultory "uncertain," and he recommended that buyers of disease management services approach the industry's vendors with "skepticism."

Well, it took him seven years, but Dr. Mattke has finally agreed with the Disease Management Care Blog that investment in disease management can be good.  Writing in the January 2014 issue of Health Affairs, Dr. Mattke and other colleagues from RAND look at the impact of disease management involving thousands of employees at PepsiCo.

In 2003, Pepsi started an employee health program that included risk assessments, on-site wellness events, lifestyle management, disease management, complex care management, telephone nurse advice lines, and maternity management. By 2011, there were 5 telephonic lifestyle programs (weight management, nutrition management, fitness, stress management and tobacco cessation) and 10 telephonic chronic disease management programs (asthma, coronary artery disease, atrial fibrillation, congestive heart failure, stroke, hyperlipidemia, hypertension, diabetes, low back pain, and chronic obstructive pulmonary disease).

Of the greater than 67,000 Pepsi employee participants, 2,610, 17,432 and 2,162 persons with an average 6.4 years of participation in disease management, lifestyle management and both, respectively, were matched, using propensity scoring, to Pepsi non-participants. The two groups' insurance claims expense and absenteeism were compared.

Overall, all the participants had an average of $360 per member per year (PMPY) less cost compared to the non-participants. The participants' vs. the non participants' cost curves diverged and became statistically significant after 3 years

However, it turned out that the savings was confined to the disease management population, which had a lower cost of $1632 PMPY.  Participants in the lifestyle management had negligible savings.  Disease management had a return on investment of $3.78

Participants in both disease management and lifestyle programs had a savings of $1,920 per year.

Despite the lack of any impact on claims expense, lifestyle management was associated with a reduction in self-reported absenteeism of .13 days per year.  In contrast, disease management had no impact on absenteeism.

The DMCB's take:

This builds on the evidence (like this and this) that later generation, remotely based telephonic disease management can reduce claims expense.  $360 PMPY for $67,000 employees translates to more than $24 million in savings per year for Pepsi. Even if the company spent millions on its health programs, the impact is something that both the employees and shareholders can be happy about.

As skeptics continue to wonder at the continuing commercial success of the disease management (now called "population health") industry, the DMCB reminds them that many other companies like Pepsi are also looking at their return on investment. They undoubtedly like what they see, but unlike Pepsi, are not taking the time or effort to publish their results.  Pepsi, in the meantime, deserves kudos for their commitment to the science of population health.

That being said, this is a company sponsored disease management program that is limited to employees and dependents.  The DMCB is less certain about the impact of these programs in typical "free range" commercial or government insurance settings.

The study isn't perfect, because there could be hidden biases.  As the authors point out, even propensity matching can't guarantee that the two populations were truly similar; since participation was voluntary, it's possible that the participants were more health conscious and that characteristic - not the disease management - is what's responsible for the observed savings.

More On "Why No One Believes the Numbers" and the Uncertainty of Measuring Return on Investment in Disease Management

Measuring ROI

In yesterday's posting on Al Lewis' book Why No One Believes the Numbers, the Disease Management Care Blog pointed out that the measurement of population health management (PHM) outcomes remains an inexact and still evolving science. While that can be a source of endless fascination for the DMCB, the inability of the industry to rustle up credible "return on investment" numbers has prompted some observers to condemn PHM as a waste of money.

The search for simple answers explains much of the appeal of this book.

According to author, one important solution is the "dummy year analysis" (DYA). This relies on repeated year-over-year measurements of utilization that use multiple comparison pairings of all patients with the condition of interest. When that's combined with a "plausibility" check list, Mr. Lewis says purchasers of the Patient Centered Medical Home (PCMH), disease management or wellness programs should be able to get a better fix on whether they saved any money. You can a sense of that perspective here.

The DMCB isn't too sure about that because a) other factors that have nothing to do with population health management can also impact utilization during and after the dummy years, making it difficult to assign an attributable ROI and b) entire health plan populations can likewise regress toward a regional or national mean.

The DMCB also sees three additional reasons why there may be less to this book's methodology than meets the eye:

1. When employers, health plans, accountable care organizations or other buyers have a list of names that have been through a care program, they typically want to understand the outcomes for the individuals on that list. If that's the case, the challenge is to find an adequate comparator that portrays what would have happened in the absence of the care program. Multiple options for identifying a parallel comparator have been used in published science for decades. That's difficult, imperfect, but not broken.  It remains an option.

2. While the book is replete with examples of "actuaries behaving badly," it is impossible to underestimate the influence of actuarial science and trending on premium rate setting, statutory accounting, and the regulation of insurance. As a result, if the actuaries say money is - or is not - being saved, health system leaders ignore their insights at their peril.

3. Isolating the impact of PCMH, disease management or wellness program out of all the other "noise" of a changing economy, evolving consumerism, benefit changes, electronic health record databases, medical advances, inflation and the news media is a function of an increasingly sophisticated and changing statistical sciences and computational technology. It's ironic, but one outcome has been a better description and measurement of the uncertainty surrounding a result.
 
To the author's credit, Why No One Believes the Numbers is not being promoted as the single best methodology that will lead PCMH, disease management and wellness programs to outcomes certainty. Rather, it is one option among many in asking whether a program had any financial impact.

Ultimately, therefore, that's why the DMCB advises that measuring outcomes in PHM - absent an ironclad methodology - comes down to using multiple approaches to triangulate on the truth. After reading Why No One Believes the Numbers, some readers may choose it as one of those approaches.

What Can Baseball Teach Us About the Return on Investment (ROI) of Disease Management? Nothing, Actually.

Growing up in New York City and then living outside of Philadelphia, you’d think that the Disease Management Care Blog would have a better appreciation for major league baseball. It likes the stadium spectacle (and food), but league standings, box scores, individual player statistics and television game play is as soporific to the DMCB as looking at the aortic valve “Mercedes Benz” sign on an echocardiogram. Try as it might DMCB cannot understand what gets baseball fans and cardiologists so atwitter.

Maybe it's because the DMCB thinks because baseball is so linear. The entire game consists of a series of singular events involving individual players surrounding one ball occurring over the time dimension of nine innings. Sure, there are other moving parts, enormous talent and high drama and but the game has a compelling degree of compact simplicity.

To each their own, says the DMCB, which is one reason why it tried to compact the moving parts, talent and financial drama of disease management (DM) into a (curvi)linear display. This is an admittedly very simplistic and not-drawn-to-scale graph of what financially happens to the return on investment in an ideally executed one-year DM contract:

The black line represents the typical accumulated savings of a well run disease management program over time. Thanks to the use of health risk assessments and predictive modeling, the patients that initially get recruited in the program are those that are a) most vulnerable and b) most amenable to care management coaching and outreach. They’re “rescued” from unnecessarily visiting emergency rooms and being unnecessarily admitted to the hospital. Note that savings grow rapidly. With time, additional patients at lower risk are recruited, but since they’re at less risk, the savings curve begins to flatten out. As more patients get recruited, it’s possible that savings can erode, because the low risk/low utilizing patients can ironically be prompted to seek out additional care services.

The red line represents the cost of the disease management program over time. Initially, there are steep and fixed start-up costs which then slow down over time but never flatten. As additional patients get recruited, more personnel and infrastructure are required for coaching and follow-up. If the program pursues each and every additional patient with multiple attempts at outreach, costs can accelerate.

The small yellow area displays when the savings exceed the cost. That means the "return on investment" (ROI) is positive. That doesn’t happen early in the program and it doesn’t happen late: it's in the middle. Early and late in the program, there are losses. Also note that the ROI is a moving target that changes that as more patients get recruited. This also explains why a positive ROI can be achieved when far less than 100% of patients are engaged in DM.

Of course, this is very rudimentary and doesn't take into account the moving parts of the baseline utilization patterns, the enormous talent of the nurse-coaches and the high drama of how costs and savings are actually calculated.

Every baseball game is arguably unique, but DM is far more complicated. It's sort of like baseball, but all the players have bats, there are 100 balls, pucks and Frisbees in play and the bases move as the game goes on. That's one of the reasons it's so much fun.

TRICARE Saves Taxpayers Millions on Chronic Illness Disease Management

Suppose you were on the leadership team of a 9.2 million member, $45 billion health insurance plan that had enrolled 80,000 persons with asthma, 11,000 with chronic heart failure and 225,000 with diabetes? Well, if you were leading TRICARE (the health insurance program for members of the military and their dependents), you would have had those persons enrolled in disease management. Fortunately for Disease Management Care Blog readers, the leadership didn't stop there: they also used standard research methodologies to assess whether the taxpayers were getting their money's worth AND submitted their results for peer review in the American Journal of Managed Care.

TRICARE has three regional “managed care support contractors” (MCSCs) that each provide disease management services (MCSCs). TRICARE forwards the names of patients with high rates of medical service utilization to the MCSCs who, in turn, contact the patients for participation in disease management. Patients can agree to receive “personalized telephonic counseling and educational mailings” or just mailings or choose to opt out entirely. If patients agree, their intervention includes a 40-50 minute baseline telephonic assessment, monthly follow-up calls to set/review care goals, additional educational mailings, newsletters and emails. Depending on the MCSC, patients can be graduated either after 12 months or after they've met their educational goals.

73,156 patients were contacted over the two years leading up to September 2008. Approximately 9000 opted out, 4700 were only in the program for 6 months and others lost TRICARE eligibility or were excluded because of end stage renal disease or HIV. That left 57,490 (about 23,000 with asthma, 4000 with heart failure and 29,000 with diabetes) for the outcomes analysis.

A control cohort using claims data from October ’04 to Sept ’05 of “comparable” patients with asthma, heart failure and diabetes populations was then fashioned to help the researchers better isolate the impact of the disease management programs. There were slight differences – the DM populations had slightly worse Charlson Comorbidity Index scores, higher rates of ambulatory visits as well as lower rates of ER and inpatient days.

As you might expect, all 6 groups of patients (the three intervention groups and the three control cohorts) experienced a pre-post decrease in claims expense. However, the patients in disease management experienced a bigger drop, which was calculated on an adjusted basis to from $152 for heart failure (which was not statistically significant) to $783 for diabetes (p < .05) to $832 for asthma (p < .05).

Patients with asthma also appeared to be more likely to get spirometry (but NOT controller medications), patients with heart failure were more likely to be on life-saving beta blocker and ACE inhibitor medications and patients with diabetes were more likely to get A1c, retinal and urine testing) – all of which were also statistically significant. Finally, patient surveys were littered with “agree” and “strongly agree” answers to surveys asking about the disease management programs’ effectiveness.

Bottom line? The total cost of disease management from September 2006 to September 2008 was $22.7 million. Cumulative gross savings was $28.5 million, with a return on investment (ROI) of 1.26. The DMCB believes that is not only statistically significant, it's quite financially significant.

While the canard that disease management “doesn’t save money” continues to stalk the hallways of academics, policy makers and regulators like a zombie that refuses to die, there is an emerging body of literature showing that disease management has evolved considerably from its early days. This study, conducted in a commercial setting involving millions of members, showed how other leaders of insurance plans can save some serious money.

Last but not least, this is a way for taxpayers to save some serious health care money. Hopefully the leadership in the Obama Administration will take note of this good news.

What Are the Top Ten Features of Cost-Saving Employer Sponsored Wellness Programs?

In the second day and concluding day of the WRG Conference, we heard from WebMD’s Larry Chapman. He echoed yesterday’s comments from Emory University: there is no doubt that employer sponsored wellness programs (many of which resemble classic disease management) have a return on investment (ROI). There are multiple positive studies from disparate settings including NORTEL, Duke University, the City of Birmingham and DuPont.

What are the lessons from such successful wellness programs you ask? Good thing the Disease Management Care Blog kept notes:

1. Don’t limit ROI economic measures to just claims expense. Include turnover, absenteeism, disability, workman’s compensation and presenteeism. That may inflate the ROI, but these domains are also important to employers and they want to know,

2. Enhance risk reduction and mitigation by promoting employee awareness, increasing motivation and helping them develop new skills,

3. Use a total population health model that is ‘results oriented,’

4. Include employees’ spouses,

5. Require employee participation in an annual health risk assessment (HRA). Think about making it part of open enrollment,

6. Let the employees fund most if not all of wellness incentives through premium differential. Don’t be shy about ‘play or pay’ and ‘getting everyone on the wellness bus,’

7. Coordinate wellness with the insurance benefit, particularly with consumer directed health plans (CDHPs) and any other cost sharing approaches,

8. Promote self care. An example is promoting WebMD’s ‘symptom checker,’

9. Promote consumer education that includes not only condition but benefit management. In fact, consider requiring CDHP participants attend a ‘how to’ workshop. Furthermore, emphasize injury prevention (an example is seat belts) and provide aggressive intervention programs for enrollees with multiple health risks,

10. Offer a tiered wellness incentive with real money based on explicitly defined criteria.

Update on the WRG Conference - A Sampling of Some Interesting Stuff

What an interesting WRG conference. For your reading pleasure, below are summaries of some of the presentations that really caught the ear of the Disease Management Care Blog. More to follow tomorrow…..

Blue Cross Blue Shield of Michigan is finding that offering wellness is an increasingly critical ingredient in winning accounts. Their approach is to offer a suite of wellness and disease management options that, depending on the buyer, can be ‘dialed up’ or ‘dialed down.’ They are developing their own assessment methodology that not only calculates savings vs. costs (i.e., return on investment) but changes in ‘net savings.’

Most interesting message: As outreach progresses from high to low risk, Michigan's modeling suggests that increasing the outreach to more persons with chronic illness eventually leads to a ‘tipping point’ decline in ROI and net savings.

StayWell believes best practice elements for company wellness programs include: strong organizational commitment, identification of wellness champions, linkage to business objectives, effective communications, having fulltime dedicated staff/vendors, making employee spouses eligible, offering comprehensiveness, raising awareness company-wide, targeting special interventions for high-risk persons, maximizing accessibility, providing incentives and utilizing biometric screening.

Most interesting message: Want to incent employees to fill out that health risk assessment or show up at a wellness program offering? Employee premium differentials beat cash rewards, which beat non cash rewards. Premium differentials not only increase participation rates, the non-participants subsidize the participants – which payers/purchasers like.

Emory University says there is a rich body of occupational health literature that has been around for years that conclusively shows wellness programs have a positive return on investment. There is a methodology using a health risk assessment that can assign a ‘Risk Profile’ to an employee-participant. This profile correlates with insurance claims expense. The ready availability of ‘propensity scoring’ makes a parallel control group readily available. Between the Risk Profile and the propensity scoring, analysts can approximate return on investment without having to conduct a randomized clinical trail.

Most interesting message: It is not uncommon for ROIs to exceed 3 to 1. And we shouldn’t be embarrassed to say so.

Highmark thinks that if you’re going to offer wellness, you might as well lead with your employees. That’s what Highmark did with an in-house wellness program. Read all about it at the February 2008 issue of the Journal of Occupational and Environmental Medicine.

Most interesting message: The ROI for a wellness program may not become apparent for three years.

Aetna was mentioned in a prior DMCB post, which discussed a blog post describing the insurer’s use of a lottery with a financial award to increase medication compliance. It turns out this is more sophisticated than just a simple lottery. There is a considerable body of research that shows humans tend to overestimate their chances in such games of chance. The medication-compliance lottery is a conscious exercise in “asymmetric paternalism,” in which the insurer harnesses their enrollees’ tendency to exercise poor judgment.

Most interesting message: Persons can be incented to make the right decision using bad decision logic. We can simultaneously harness and respect a person’s right to choose - wrongly.

Return on Investment, Disease Management and Wellness

The Disease Management Care Blog is coming to you from Washington DC, where it is attending (and speaking at) a World Research Group conference focused on wellness. It intends to share the highlights of the other speakers from other settings who are embarking on new programs in future posts, so stay tuned.

One of the major themes of this confab is the perennially difficult topic of ‘return on investment.’

Some of the DMCB’s planned – and simplistic - comments for tomorrow:

If we must use the term ‘return on investment,’ its measure is generally done one of five ways. These are:

1. Compare the claims expense of either the population or a representative part of the population to a matched control. Using the population at baseline and comparing it to itself (pre-post) after the intervention is a variation on this theme. It’s better to use a parallel control. This has the advantage of being conceptually easy to understand. This may partially explain why this approach still appears, in the experience of the DMCB, to dominate the market place.

2. Examine the trend (change over time or the slope) in claims expense for the population and compare it to a control or what the expected trend should be. Trend can be more difficult to understand but it has the advantage of also being used in other health insurance calculations. As such, it is probably destined to be the ‘coin of the realm.’

3. Use ‘Other Weird Calculations’ such as measuring the relative impact of the program on the observed trend. In other words, as the claims expense goes down (or up), is there any correlation with the amount of the intervention, and if so, how much? For example, do more coaching calls translate to correspondingly fewer admissions?

4. Use anecdotes. Don’t underestimate the impact of positive or negative personal testimonials shared at an all-employee meeting or given to the head of Human Resources.

5. Rely on Quality Adjusted Life Years (QALYs), which captures the possibility that there isn’t a reduction in claims expense. If there aren’t, what gains in quality are there, and for each ‘unit’ of quality what is the cost?

The DMCB will be looking forward to hearing how others at the Conference tackle this. More to follow.