How Can We Align Financing Incentives With Quality?

Dennis P. Scanlon, PhD: I wanted to ask about scientific evidence in this area of this disease. There’s preapproval science and there’s [the information] we learn in the postapproval real-world implementation phase. We’ve identified some risks over the years that the studies didn’t completely elaborate on.

We have some producers of research here, certainly consumers, and an editor of a journal. What are your thoughts on the science and the process of developing science? Do you have any thoughts in terms of how we might need to do that better or where we might need some changes?

Zachary T. Bloomgarden, MD, MACE: One very important thing to look at is the inclusion and exclusion criteria of every trial and make sure that the trial participants really resemble the person that you’re sitting in front of—who you want to choose a medicine for.

There was an analysis of this Scottish diabetes database which looked at all of these large cardiovascular trials—going from PROactive (Prospective Pioglitazone Clinical Trial in Macrovascular Events), to ACCORD (Action to Control Cardiovascular Risk in Diabetes), and so on. They found that if you look at how representative the trials are, they go from about 3% to, at most, 30% of people with diabetes in that community. That’s a big dilemma.

So, either two-thirds, or more than 95%, aren’t going to fit the characteristics of the people studied. That, absolutely, is going to apply to these cardiovascular safety trials. So, before we say everybody should take empagliflozin, you might want to say, “Everybody who resembles those 6000 people who were carefully selected to have a certain level of risk factors [should take empagliflozin].”

On the other hand, if we look at the larger epidemiologic databases and say, “Those individuals who took DPP-4s (dipeptidyl peptidase 4s) have better cardiovascular outcomes in the community than those who take sulfonylureas,” do we really know that there’s not some bias that led physicians to prescribe DPP-4s to individuals with different socioeconomic, educational, or other characteristics from those who got a sulfonylurea? If so, those may be the real causes of the better outcome with the DPP-4. So, there are a lot of questions, and I don’t know that we’ll ever get the answers.

Robert Gabbay, MD, PhD, FACP: There’s a limit, I think, as to how much we can “data mine” from retrospective data [and determine] that something is safe or more effective. It’s hypothesis-generating and helpful, and, certainly, there have been numerous examples where signals or problems were identified through [analyzing] large data sets. But, it can’t be definitive. For better or worse, there have been seismic changes in the way we’re thinking about managing our patients because of these trials.

Dennis P. Scanlon, PhD: But, it’s important to do some post monitoring?

Robert Gabbay, MD, PhD, FACP: Absolutely.

Dennis P. Scanlon, PhD: For safety, especially.

Zachary T. Bloomgarden, MD, MACE: We also have to be careful not to overreact to small signals. Recently, there have been a few reports that suggest that DPP-4 inhibitors might be associated with a kind of arthralgia syndrome. Millions of people have taken these drugs; 33 cases were reported to the FDA, and it sounds like those people did have a reaction.

I have 33 people in my office every week who have terrible arthritic symptoms. How did those few cases lead to a warning label on every product information for a DPP-4 inhibitor? We have to do it the right way.

Michael Gardner, MD: The same could be said for DKA (diabetic ketoacidosis) with the SGLT2s (sodium-glucose cotransporter type 2s). When you really start to parse down the data, is it really all that unexpected? Is it any different than the comparators? It’s really beginning to look like it may not be.

Zachary T. Bloomgarden, MD, MACE: We had an AACE (American Association of Clinical Endocrinologists) conference on that whole phenomenon, and there’s some fascinating science behind it. It turns out that, for utterly unexpected reasons of SGLT2’s role in the alpha cell and effects on tubular reabsorption of ketone bodies, this really may be an issue.

Dennis P. Scanlon, PhD: Why don’t we move on and talk about quality measurement, quality metrics, and so forth. Anytime we think about quality metrics, HEDIS (Healthcare Effectiveness Data and Information Set)-type measures and such, people often point out patient compliance and variation across patients in terms of their behaviors.

Diabetes is an area where willingness to adopt a healthier lifestyle, to exercise, and to diet can significantly make a difference. From your perspective, maybe not only in choosing a therapy and a treatment for a patient, but also in thinking about this from a population perspective, what is the impact on the quality measure?

Robert Gabbay, MD, PhD, FACP: Number one: I think we need more nuanced quality measures that take other things into account. But, in the absence of that, I think in all of the quality measures and things that we see from managed care and elsewhere, the goal is not that 100% of patients should meet any goal. [Rather], there’s an exception for a percentage of patients based on individualized needs, cases, and comorbidities, those who are not ideally suited for that therapy.

The other thing I’m always a little nervous about, as a healthcare community, is us saying, “Well, it’s all about patient adherence, and that’s not my problem.” We have an important role in that, and there’s certainly [applying] interventions that we know can improve adherence and engagement of patients. It’s part of our duty as providers to be able to guide people to do the things that will be better for their health.

Zachary T. Bloomgarden, MD, MACE: I agree with what you’re saying. One of my arguments with the American Diabetes Association’s latest guidelines is that they show many dimensions of decision making as to how low or high an individual’s A1C (glycated hemoglobin) goal should be. One of them is the person’s willingness to treat their diabetes as it were—their adherence level. But, we could use that and say that for anyone who is, perhaps, not sufficiently motivated because of extrinsic factors, that’s not someone we’re going to try to treat [as intensively]. Yet, that’s not fair to that individual.

If a person has to commute an hour and 15 minutes each way on the New York City subways to get to work and doesn’t have time to bring healthy food with them, but instead has to get something at the local deli and then has no time to exercise, should we say, “Well, this is someone who we’re just not going to try to work on lifestyle with?” Or should we [ask], “How can we reengineer their life environment to help him or her to be able to do this?”

John A. Johnson, MD, MBA: I think the important thing with the quality measures is that they are a proxy for helping us manage the outcomes.

Robert Gabbay, MD, PhD, FACP: And the overall population.

John A. Johnson, MD, MBA: Exactly. If we have a better look into, from a population health perspective, those patients that have a diabetic retinal exam Star or have a hemoglobin A1C (HbA1C), and they’re controlled, from all the studies that we’ve referenced, we know that we can use that data to predict, or at least manage, the downstream effects from poor quality better. That’s really where the managed care population steps in and likes to partner with the provider and specialist to [ask], “How can we improve the quality such that we can control some of the downstream costs?”

Dennis P. Scanlon, PhD: Your Medicare Advantage plans obviously have a market basket of quality indicators that aggregate up into Star Ratings, which have a fairly significant financial impact depending on how you do. Diabetes, of course, is just one of many conditions reflected in those. Any thoughts on the relative weight that diabetes gets, in general, in the determination of those plan ratings, or the specific quality measures?

John A. Johnson, MD, MBA: If you look at a Star Rating, again, the biggest components of stars would be HEDIS and CAHPS (Consumer Assessment of Healthcare Providers and Systems). CAHPS is the consumer survey that the general public, if they’re included in the sample, fills out and gives an assessment on [of] the care they receive from the health plan, whether they have diabetes, and the care they receive from their provider. HEDIS gives us more quality data on how the provider, the patient, and the member is performing.

Diabetes is high because some of the measures are triple-weighted: controlling blood pressure, HbA1C, and medication adherence—not just that the doctor did a great job of prescribing it, but ensuring that the person is actually taking it and getting it filled. All of that weighs heavily into the assignment of quality performance and the Star Ratings.

Dennis P. Scanlon, PhD: I assume [though] that the medication adherence is subject to the discussion we just had, which is within a sort-of endogenous choice situation around what is best.

Zachary T. Bloomgarden, MD, MACE: Controlling for the severity of illness is so important, [as well]. Endocrinologists tend to see individuals who have the higher level of illness—who have the higher A1C. There have been some earlier analyses which said that endocrinologists have worse outcomes than primary care. Sure—if they start with much more ill patients.