What happened?  Disease management can be a real help to people with chronic illness, and it can save claim dollars for employers and other payers.  That promise created a booming market for disease management, and a gold rush ensued as new entrants flooded the market and eroded the value of the service.  Suddenly every health plan offered a disease management program, and many did not bother building the solid nursing foundation needed to deliver real results.

The heart of disease management is a therapeutic relationship between a nurse and a patient.  That’s how you educate, change behaviors, assure quality care and create better outcomes.  It helps if you can enlist the patient’s physician as well.  Boxes full of literature, glitzy marketing campaigns and grandiose claims don’t do the difficult jobs of establishing that relationship, supporting patient education and self-care, and delivering better results.

Nurses are expensive.  If you are a disease management company, you can under price your competition by having fewer patients in nurse coaching.  And then don’t bother reporting that to the client – read a disease management activity report today and try to figure out how many health plan beneficiaries are actually getting nurse coaching.  It’s the critical metric, but most disease management vendors don’t report it to their clients.

What’s the right number of a health plan’s diabetics that should be in nurse coaching?  50%?  25%?  We can debate that, but I will guarantee you it’s not less than 2%, which is the result for some disease management programs today.  There is no possibility that touching so few beneficiaries will result in a meaningful impact or real program savings.  Unfortunately, disease management vendor estimates of savings and ROI are so mired in statistical problems, including selection bias and regression to the mean, that they cannot be relied on as a meaningful measure of the program impact.  The industry missed a chance to standardize measurements of results to report valid metrics and allow meaningful comparison across vendors.  This would have created a fair marketplace, assured accountability, and rewarded top performing organizations.  Accountability and transparency could have headed off the problem we are approaching when nobody believes disease management savings and ROI reports.

We know we have a problem as disease management vendors withdraw from the Medicare demonstration project and reports appear more frequently that question the value of disease management.  There’s a chance that this is simply part of a healthy shake-out in the marketplace, but those of us who see the real promise of disease management also fear that the coming backlash will go too far.

In the first three parts we’ve discussed the problems of finding all the potential candidates with disease management conditions among an employer’s beneficiaries, of getting coaching to a significant number of those identified candidates, and of demonstrating the financial impact of the program to justify the dollars being spent.  While these are the three biggest barriers to having an effective program and getting meaningful results, there are lots of secondary issues which can prevent an employer from realizing the potential value of a disease management program:

1. Clinical measures of improvement – we have noted the difficulty of calculating a valid return on investment and the delay of 18 months or so between the time that an employer starts paying for disease management services and when that employer can get a calculation of savings.  It is critical, therefore, to look at intermediate measures, and there are excellent ways to demonstrate the clinical improvement of the participants.  Diabetics, for example, should show improvement in their hemoglobin A1c levels as control of their blood sugar improves, should show better compliance with medications, and should have recommended testing for cholesterol, kidney function, and eye damage.  Improvement in these measures will predictably and reliably result in lower claims costs and better health outcomes.  Conversely, if there is no improvement in these measures, than any demonstrated “savings” will not be real.  Yet many disease management vendors omit this information or deliver it in a way that does not allow meaningful validation of their services.  If you don’t see a clinical improvement, don’t believe a vendors claim that they are saving money.
2. Failure to incorporate pharmacy data – disease management programs that rely only on medical claims information will not find most potential candidates and will overstate the savings generated by their programs.  Physician claims are notoriously inaccurate, since providers fill out forms to get reimbursed, not to provide accurate information.  Hospital claims are more valid, but they will identify only the sickest of the sick, leading to extreme regression to the mean and hugely inflated savings when it comes time to report return on investment.  Pharmacy data is essential to identify beneficiaries with asthma and diabetes, and to find heart disease and other diagnoses at an early stage.  Also, medication compliance is a critical element for controlling long-term costs, and if you are not tracking prescription utilization, there is no valid way to assess compliance.  If your disease management program does not incorporate timely pharmacy data, you will not have an effective program, although the return on investment report will look great.
3. Encouraging participation.  This is a difficult issue, because responsibility for getting identified candidates to accept coaching is shared between the employer and the program.  If employees are not familiar with the disease management program, they will be unwilling to participate, so employers must publicize the introduction of the program and assure employees that their medical information will not be available to the employer.  But even with good communication, only 20-30% of participants offered disease management coaching will accept it, in the absence of incentives.  Fortunately, sicker beneficiaries are more likely to accept coaching, and a sharp disease management program will make a second offer to a candidate who initially declined participation after that candidate has had a significant event like an ER visit or hospitalization.  But providing an incentive for participation, most often a modest refund of payroll contributions or other financial reward, will increase participation and improve program effectiveness.  Structuring such contributions to avoid HIPAA concerns and allow ease of administration can be challenging.
4. Coordination with other health plan programs.  Particularly when an employer contracts with a stand-alone disease management company, there can be conflicts between the disease management coaching and the health plan’s patient programs.  Case management is the most significant example, where beneficiaries with catastrophic illness like organ transplantation, cancer, and extensive rehabilitation have an intensive relationship with a health plan nurse.  It is clear that the nursing directives for such overwhelming illness must take precedence over routine coaching for diabetes or asthma, but often there is no connection between the clinical staffs at the two organizations to coordinate instruction and care.  Patients may be left confused, frustrated, and unwilling to participate in either program.  Similar conflicts can also arise with health plan activities including health advocacy, 24 hour nurse lines, and hospital preadmission and postadmission counseling.  It is an important and challenging to assure that the clinical staff delivering disease management services coordinates with other health plan services.

With all these ways that a disease management program can go wrong, it may be tempting to simply forget the whole thing.  That would be a mistake, given the potential for these programs to reduce medical claims, improve disability and productivity, and produce long-term health status improvement.  Good programs are available and vendors can be help accountable for delivering real results.  It is worth the effort.

Okay, your disease management program has been running for a year or so.  You made sure your health plan beneficiaries with diabetes, heart disease and other conditions were identified early, and then made sure that they were actually getting disease management coaching by nurses, instead of “participating” through passive mailings.  You’ve paid for a lot of disease management, and now you’d like to know if the program is making a difference.  Is it worth what it cost you?

About a year and a half after the program started, you’ll finally get a Return on Investment (ROI) report.  And look – you’ve got a return of $3 for every $1 spent, maybe 5:1 or even 12:1.  It’s probably highest for asthma, but even the 3:1 ROI for COPD looks pretty good.  When you talk to another company that used a different disease management company and they’re only getting a 1.3:1 ROI their first year, you feel pretty good about your results. 

But there’s this lingering feeling that maybe those results are a little too good to be true.  You check out your asthma costs from one year to the next, and are still going up.  You look at hospitalization rates for those conditions, and they haven’t changed.  And your health plan claims haven’t gone down, even though chronic illness in general, and cardiac disease in particular, is your major cost driver.  What gives? 

Hold that thought.  Let’s try a mind experiment.  Suppose I looked at your health plan claimants, and identified everyone who was hospitalized during one year with heart disease.  Suppose I then told you I waved a magic wand over those health plan members, and the following year their medical claims were lower.  Would you be willing to pay me to wave my magic wand again the next year?

No, because you know that chronic illness is cyclical, and someone who has a hospitalization one year will not likely be hospitalized again the next.  I haven’t done anything to change those who were hospitalized in our mind experiment by waving a wand, and their costs would have been lower the following year, regardless.  It’s the same with someone who bowls 300 one night or throws five touchdown passes in a game.  The next time out they will probably not exceed their average.

That observation is called “Regression to the Mean,” and it’s the problem with much disease management reporting.  Disease management companies most often estimate their savings by picking patients who were very sick with an illness in the "baseline" year, and then measure the results the following year when they were participating in the program and were probably not as sick.  Just because costs went down does not mean the program was successful – you have to carefully adjust the results to account for regression to the mean, and preferably to compare the beneficiaries with access to the program to those who did not.  There are also other ways to make sure the calculations are fair, and other measure to assure that the program worked as promised. 

Disease management programs that are good and effective will demonstrate a return of 1.5 to 1 or even 2:1 the first year, but anything higher is likely due to regression to the mean, and not to any real improvement in the health status of your beneficiaries.  If it sounds too good to be true, it probably is.   

What is Lifestyle-Based Analytics? It is the analysis of lifestyle-based data, which is widely available through consumer databases. The analysis of this data creates enormous opportunities and advances strategies for detected lifestyle-based diseases before they escalate into advanced stages.

Where is this data found? The data is coming from a diverse group of data sets. Every time you use a credit card, swipe a discount card at the grocery store, transact business on the Internet, apply for a mortgage, or go to the health club, information about your consumer habits are captured in a database. Culling through these databases can deliver significant data like food purchases, fitness activities, stress indicators, family size, occupation, tobacco preferences, alcohol preferences, travel destinations, and vehicle preferences. This data, when aggregated, can be insightful to health and life insurers in predicting future risk.

How are Lifestyle-Based Analytics being used? One of the most common uses of lifestyle-based analytics is to augment the underwriting process. Currently, data is available that accurately reflects the past medical conditions of an individual. Coupling this process with Lifestyle-Based Analytics can assist underwriters in predicting non-hereditary diseases and give an underwriter access to information those individuals who are unhealthy, as well as on those that data sets that show practice lifestyle choices considered to be healthy.

Disk storage per Person (DSP) is the approximate measure of the average amount of information stored for any particular individual. In 1985 DSP was estimated at 0.02, in 2005 DSP is projected to be 3,500 per individual. There is strong momentum toward using this data to progress understanding of risk. With lifestyle-based diseases accounting for three quarters of the nation’s medical expense, it is difficult to see how this momentum could be shifted. A better question is, should it be shifted? 

As we discussed in Part 1, the first challenge in getting good disease management results is to make sure that your vendor has identified your health plan beneficiaries who would be good candidates for disease management programs and assessed their needs, risks, and motivation (see Part 1).   The 300 diabetic members in your health plan, for example, are getting pharmacy and physician services, and you expect to see them identified by the disease management program.

So far, so good.  Sometime later, probably after the 1st Quarter program activity is reported, you wonder how many of those 300 diabetics are being served by the program.  The answer, clearly stated in the disease management report, is that 285 members, 95% of those identified, are “participants” in the program.  Sounds great, right?  But how come none of your coworkers who have diabetes are getting the coaching that the program promised when you signed up?

As a result of “opt-out” program designs, once someone is identified as having a disease like diabetes, they are mailed some program materials and told to call to “opt out” of the program.  Less than 5% will do so – if you didn’t want to participate, would you bother to call in response to an unsolicited letter?  The result is that 95% got a letter and some program materials, did not call in to refuse the program, and are now “participants.”

The problem is that mailing some program materials does not change health behaviors, does not teach diabetics better self-care, and does not provide oversight to make sure they are getting the best possible care.  If that were all it took, we could skip the coaching entirely:  nurses are expensive.  But the evidence shows that mail-only programs do not result in better behavior or understanding, do not improve care or outcomes, and do not result in lower costs.  As you look deeper into the reporting, you realize that only 10 or 20 of your diabetics are actually getting disease management.  Yes, as few as 5% of your diabetics, or even less with some vendors, will actually get the disease management you paid for.

What you were promised when you bought the disease management program, and what has been show to be effective in improving outcomes and reducing costs, is active coaching by nurses.  It is the therapeutic relationship between a nurse and a patient that works the magic in disease management programs, and this doesn’t happen with a booklet in the mail.

Bottom line: look at the number of your beneficiaries getting active coaching, not the number of program “participants,” to see whether you’re getting the disease management program you paid for.

Disease management programs can work wonders to improve the health status of individuals with diabetes, heart disease, and other medical conditions, and to lower the employer’s cost for medical care.  Basically, whatever the diagnosis, the more you know and the better you care for yourself, the less expensive care is needed and the better the outcomes.  This is especially true for avoiding ER visits and hospitalizations.

But all disease management programs are not alike, and one recent problem has been that the wrong patients get identified.  Basically, health plans and disease management companies take medical claims and pharmacy data and try to figure out who has a medical diagnosis like diabetes.  Once they are identified, the plan calls the individual to confirm that they really have diabetes, and then “stratifies” them, basically figuring out who needs more help.

To save money, though, some disease management vendors don’t bother calling everyone they have identified, they just “stratify” based on how much medical care they have received.  The problem?  Two – claims are often coded by physicians to maximize reimbursement, not necessarily to provide an accurate diagnosis.  So as many as 10%, even 20%, of those identified as having a medical diagnosis, may not actually have it.

Even worse, if you don’t call, you have no idea what the patient’s history is, or how much they understand about their illness.  It is impossible to tell who needs help the most, without knowing this information.

Bottom line:  check if your disease management company calls everyone they identify with a diagnosis to give the best care, or is simply cutting corners.