Dr Karmacharya reviews his presentation from ACR Convergence a population-based study from the UK THIN database across psoriatic arthritis, psoriasis, and axial spondyloarthritis to find distinct and remarkably stable patterns of multimorbidity. 

Paras Karmacharya, MD, is assistant professor at Vanderbilt University Medical Center, and director of the Psoriatic Arthritis and Spondyloarthritis Center.

CLINICAL PRACTICE SUMMARY:

• A population-based UK study conducted by Vanderbilt University analyzed more than 5,000 patients with psoriatic arthritis, over 73,000 with psoriasis, and about 2,400 with axial spondyloarthritis using K-median clustering over at least one year of follow-up to identify stable multimorbidity clusters.

• Each disease demonstrated distinct and reproducible patterns: psoriatic arthritis showed chronic pain and psychiatric clusters, psoriasis showed asthma/allergy and psychiatric clusters, and axial spondyloarthritis showed osteoarthritis and high cardiometabolic/cancer/COPD clusters, which remained consistent from year one to year five.

• Patients with five or more chronic conditions had twice as many outpatient visits, 2.5 times more hospitalizations, and three times more prescriptions, indicating that the total number of comorbidities—rather than the cluster type—was the key predictor of health care utilization.

TRANSCRIPT:

Hi everyone. I'm Paras Karmacharya, assistant professor and physician scientist at Vanderbilt University Medical Center, where I also direct the Psoriatic Arthritis and Spondyloarthritis Center. In our population-based study from the UK THIN database across psoriatic arthritis, psoriasis, and axial spondyloarthritis, we discovered that patients fall into a few distinct and remarkably stable patterns of multimorbidity. And each disease seems to have its own multimorbidity cluster pattern. But when it comes to health care use, the thing that seems to matter most is how many conditions you have. And in short, count beats cluster when predicting health care burden is what we found.

Let me talk you through this, get you through this study here. So we all know that patients with these inflammatory diseases often have multiple comorbidities, cardiovascular risk factors, depression, metabolic issues, pain and more. But most studies just tally the number of conditions.

So what we wanted to know was do certain combinations occur together in meaningful ways? And if so, do these multimorbidity patterns. So multimorbidity presence of two or more conditions, chronic conditions in the same patients influence how much health care people use. And so to answer this, we use the UK THIN database or large population based primary care dataset. We identified over 5000 patients with psoriatic arthritis, more than 73,000 patients with psoriasis, and about 2400 patients with axial spondyloarthritis, each with at least a year of follow-up. And we also included 5 mass controls for each of these cohorts.

For each patient, we captured clinically relevant chronic conditions based on previous literature, things like obesity, diabetes, hypertension, depression, and fibromyalgia. Then we let the data tell us the story. So using a tree-based approach to decide how many groups existed, followed by clustering method called K median clustering, we identified clusters of patients with shared comorbidity profiles, and we found that each disease produced a distinct and stable pattern of multimorbidity clusters.

So in psoriatic arthritis, we found 4 groups: namely cardiometabolic, those with local comorbidity, chronic pain—this chronic pain cluster was unique to psoriatic arthritis—and also a psychiatric cluster which had more patients with depression. And again, in psoriasis we found 4 different clusters, which was specific to psoriasis, so local morbidity, and there was an allergy and asthma cluster, which was unique to psoriasis, and there was a back pain cluster and a psychiatric cluster, so slightly different compared to psoriatic arthritis.

And in axial spondyloarthritis as well, we found 4 different clusters: cardiovascular; osteoarthritis, which was unique to this group; and a high multimorbidity cluster marked by severe cardiometabolic disease, cancer, depression, COPD; and then there was a low comorbidity cluster as well. So what was fascinating was that these clusters were not random. They were replicable, stable over time. So to make sure that we found same kind of clusters, we looked at year 1 and year 5 after diagnosis.

And it was also, and they were remarkably similar at year 1, and year 5, and they also tended to be disease specific even within these related diseases. So in other words, psoriatic arthritis tends to bring like pain and depression, and psoriasis brings allergy and asthma, and axial spondyloarthritis brings osteoarthritis.

And next we looked at health care use and among these clusters and among the number of comorbidities in these different conditions. And so we found that regardless of the specific cluster, the pattern was pretty clear that the more comorbidities a patient had, the higher their health care utilization was. For example, in psoriasis, those with 5 or more chronic conditions, which we define as substantial multimorbidity, had 2 times more outpatient visits, 2 1/2 times more hospitalizations, and 3 times more prescriptions used than those with low comorbidity and in psoriatic arthritis patients in the chronic pain cluster had some of the highest rates of hospitalizations and prescriptions.

And in axial spondyloarthritis, the high multimorbidity cluster, those with cardiometabolic disease plus cancer, depression, COPD, had nearly 3-fold higher prescription rates. So while the patterns help us understand disease heterogeneity, and these seem to be somewhat disease specific, from a practical standpoint, it seems like the burden, not the type of multimorbidity, seems to drive at least the health care use.

So this tells us 3 important things. So in clinical practice, we should be thinking beyond the primary diagnosis. So a simple morbidity and count, even something like 5 or more chronic conditions can be a powerful signal for patients and that may need closer monitoring integrated care or social work support. And then secondly, certain patterns can help us anticipate specific needs. For instance, chronic pain management and mental health integration might especially be relevant to psoriatic arthritis. And then asthma screening could help some patients with psoriasis, while osteoarthritis seems to have a higher burden in axial spondyloarthritis and prioritized in this condition.

So third, from a system perspective, we should probably focus on shared drivers—cardiometabolic disease, depression, pain, these cut across all 3 conditions and likely account for a large chunk of the axis utilizations we see. And so looking at, of course, this is an observational study limited by quoting accuracy and unmeasured factors as well. But the consistency of clusters gives us some confidence that these patterns are real and clinically relevant. And the next step would include embedding simple burden screens, for example, flagging patients with 5 or more chronic conditions into routine care, and then testing whether cluster-informed care bundles can actually reduce utilization and improve outcomes in these patients.

So to summarize, each of these diseases seem to have their own multimorbidity pattern, but the driver of health care use seems to be mostly the total number of comorbidities and so its important to recognize the cluster, but act on the count. Thank you very much.