Conversations about healthcare for people with Alzheimers Disease and other dementias (ADOD) often center around finding drugs that can prevent, slow or reverse the devastating neurologic symptoms. And, of course, that is critically important. But also important is managing the overall health of people with dementia. This includes optimizing care for their comorbid chronic conditions, such as heart disease and diabetes as well as preventing medical complications related to dementia, such as falls, fractures, urinary tract infections, and pneumonia.
While consulting with a novel startup, Ceresti Health, that is developing digital and human capabilities to empower caregivers of patients with dementia to be better able to help with home management of their loved ones’ chronic conditions, I became interested in the question: What is the impact of dementia on the costs of modifiable comorbid conditions?
Cost of Care
My interest was spurred because I knew, after working in the industry for decades, if there is one thing that motivates payers, at least in the U.S., it is the prospect of significantly reducing the cost of care of their covered lives. It has been estimated that the expenditures for total health care and long-term care payments for Medicare and Medicaid’s patients with dementia were ~$159 billion in 2017. This number is expected to exceed $1 Trillion by 2050.
We know that a great deal of this money is spent to care for comorbid conditions as opposed to the treatment of underlying dementia. I wondered, how much of that $159 billion could be saved NOW if there was better management. Not just of dementia, but of the comorbid conditions, like diabetes and heart disease, that we already know how to manage and for which we already have relatively inexpensive, effective drugs to do just that.
To answer the question, we pulled together a team comprised of an independent epidemiologist with years of experience in the field of population health, the former CEO of a large population health company and current Ceresti part-time executive, Ceresti’s CEO, myself (an occasional consultant to Ceresti).
The results of our research were just published in the American Journal of Managed Care*[registration but not payment is required to view full text]. AJMC is a peer-review journal that publishes research related to issues important to stakeholders in the managed care space.
We performed a cross-sectional retrospective analysis of 2010 Medicare Claims from more than a million FFS Medicare beneficiaries (of note, the data set did not include drug claims). We identified a little less than 100K beneficiaries that had at least one claim (ICD 290, 294, and 331) for Alzheimer’s or other dementias – these folks are labeled ADOD in the tables below.
Because of the substantial impact of age in particular on the prevalence of Alzheimer’s, we used propensity score matching to adjust for age, sex, and race. An important design element in our study that differentiates us from most other dementia cost studies is that we purposely did not match for the comorbid conditions. This is because we wanted to study the impact dementia on the cost of those conditions.
Using a modified Delphi panel, we selected 15 comorbid conditions that are potentially modifiable by care management to study. We did not include mental health comorbidities because of difficulties with diagnosis and treatment in the setting of dementia.
The Number of Diagnostic Fields is Important
An important take away from our study design is that we looked for the dementia codes in all 10 diagnostic fields reported in our data set. We found that we would have only identified 35% of dementia cases if only the first field had been examined. In fact, the first 4 fields were required to get to 80% of the cases, leaving 20% unidentified as having dementia.
This is an important detail because most organizations that use claims to calculate prevalence only use two or three fields instead of all ten. And, dementia is often not listed amongst the first 2-3 fields because doctors always list the most immediate cause of an office visit or hospitalization (e.g., heart attack or COPD exacerbation) first, noting the presence of dementia much farther down the list of more acute (or some might say more billable) conditions.
ADOD is associated with increased cost of care.
Our analysis revealed that 9.4% of the total population had dementia but they accounted for 22.8% of costs. The converse is that 90.6% of the beneficiaries who did not have dementia accounted for only 77.2% of the costs. Remember, these data are from the total sample and are therefore confounded by significant differences in age, race, and gender.
After we propensity score matched the population, we ended up with two equal-sized groups each with 99,483 beneficiaries. Now, as you can see in Table 1 below, 50% of the matched sample had dementia and 50% did not. The new groups were well matched for age, race, and gender, but not matched for the chronic conditions that we were studying.
Look what happens to the distribution of costs. They have almost flipped with beneficiaries with dementia now accounting for almost 70% of the costs and those without dementia, only 30%. Comparing the mean and median costs of individual beneficiaries, we found that the costs of those with dementia remain the same in the matched and unmatched groups. This is expected because they are exactly the same people. Individual costs in the groups without dementia increased only slightly perhaps reflecting the impact of matching for age, sex, and race.
It was our hypothesis that the dramatic difference in the cost of care was due, in part, to the impact of dementia on the management of these comorbid conditions.
Prevalence of the conditions is greater in the ADOD group.
Further analyzing the matched groups, we found that the prevalence of all of the study conditions was greater in the dementia group compared to those without dementia (Table 2). To make this easier to understand, we calculated the prevalence ratio (PR) by dividing the number of people with dementia by condition by the number of people without dementia with the same condition. A number greater than one indicates that the prevalence of the condition was higher in the dementia group compared to the group without dementia. The PR was greater than one for all 15 study conditions.
When we looked at how many beneficiaries with dementia did not have one of the 15 study conditions, it was only 19% in the dementia group compared to 36% in the non-dementia group. This is an important finding when because some of these study conditions are used as exclusion criteria in some Alzheimer’s studies.
The impact of dementia on each comorbid condition
In order to understand the impact of dementia on each comorbid condition at the level of the individual beneficiary, we looked at the costs of individuals with and without dementia who had only one of each the study comorbidities. (Of course, we could not control for other comorbidities or conditions that we did not study.)
Table 3 shows what we found. For each study comorbidity, the mean annual cost for individuals was substantially higher in the dementia group with costs differences ranging from a low of $2,738 per person annually for influenza (not shown in Table 3) to a high of $8,503 per person for UTI.
Therefore, not only were the chronic conditions more prevalent in beneficiaries with dementia, but each of the chronic conditions appeared to be more costly in people with dementia compared to those without dementia.
This supports our hypothesis that chronic conditions may be more costly in patients with dementia compared to those without dementia. It is possible that this is because the management of many of these chronic conditions has an important component of self-care that may be impaired if the beneficiary also has dementia. It also could be that clinicians caring for these patients were less aggressive in recommending treatments. Or that patients with dementia had trouble accessing care due to mobility limitations. We also felt that some of the conditions may be related to dementia itself; for example, falls or UTIs. But even these conditions are potentially modifiable with improved individual care and population management.
Guidance for health plans and other organizations that manage care
In order to provide guidance to health plans and other organizations that manage care, we attempted to prioritize each of the study comorbidities at both the individual and population health level. To do this, we calculated what we called the Individual Cost Ratio or ICR (the difference in cost for each condition with and without dementia divided by a standard–-in this case we used a simple mean of the individual differences).
What we found (Table 4) was that the single comorbidity with the highest individual cost ratio (1.5) was UTI. Individuals with dementia and only UTI had $8,503 higher average costs compared to individuals with only UTI in the population without dementia. Diabetes with complications ranked second with an ICR of 1.4 and $8,092 higher mean costs. And fractures ranked 3 with an ICR of 1.3 and $7,260.
We also calculated a Population Cost Ratio (PCR) that takes prevalence of the condition into account. The prevalence of a condition is an important consideration for health plans and other organizations with population health programs as they try to decide how to allocate their resources.
We calculated the PRC by multiplying the ICR by the number of patients with dementia and then dividing by 10,000 so that the numbers would align with other numbers in our study. As you can see in Table 4, we found the study comorbid condition with the highest PCR was the fracture group (PCR 5.7) followed closely by UTI (5.6) and then Uncomplicated diabetes was a distant third with a PCR (2.4).
Alzheimer’s disease and other dementias have a substantial impact on both the prevalence and costs of certain comorbid conditions that may be modifiable by care management and population management.
These are conditions that we already know how to manage or prevent and for which we already have effective, relatively inexpensive therapies.
We believe if these therapies are optimally applied, there is a potential to save a huge amount of money. In fact, multiplying the per person difference in costs by the estimated prevalence of Alzheimer’s in Medicare, it amounts to an estimated $60 billion dollars per year (in 2010 dollars) some of which could potentially be saved by optimally managing these comorbid conditions. This is money that could be repurposed to help find a cure for the disease.
Of course, future studies will need to be performed to determine if more intensive management of these comorbid conditions will actually significantly alter the costs of care for Medicare beneficiaries with dementia and result in the savings. They will also need to determine how large the savings are after subtracting the increased costs related to improved management.
Finally, our study strongly suggests that clinicians, caregivers, plans and other payers, like Medicare should not only focus on managing the patients’ dementia, but also on optimizing the care of their other comorbid conditions as well. And, we believe, that in addition to continuing the quest for a cure, researchers and research dollars should also be dedicated to finding ways to improve care and outcomes for these serious, expensive modifiable comorbid conditions.
*The original paper, “Impact of Dementia on Costs of Modifiable Comorbid Conditions” was authored by Patricia R. Salber, MD, MBA; Christobel E. Selecky, MA; Dirk Soenksen, MS, MBA; and Thomas Wilson, PhD, DrPH. It was published in the American Journal of Managed Care on November 8, 2018. It can be found here: http://bit.ly/2zLfcck