GeneiData Intelligence Lab models predict use of specialty drugs, Humira and TNF inhibitors.
Data Science

GDI Lab model to predict specialty drug use

September 9, 2021
Predicting the use of Humira and TNF inhibitors in RA patients
Principal Data Scientist


Chances are you’ve seen a television or print advertisement for Humira. That’s because AbbVie spent an estimated $45 million in advertising in December 2020, up from $40 million in November 2020.

Humira ad

Humira, the brand name of AbbVie’s adalimumab, is prescribed to “reduce the signs and symptoms of moderate to severe rheumatoid arthritis (RA) in adults.”

Allow me, as a registered pharmacist, to share additional details about Humira.

Humira is a type of tumor necrosis factor (TNF) blocker. TNF blockers “suppress the immune system by blocking the activity of TNF, a substance in the body that can cause inflammation and lead to immune-system diseases.” The use of Humira or another type of TNF blocker suggests a person’s RA has deteriorated.

Humira also belongs to the drug type called biologic disease-modifying anti-rheumatic drugs (DMARDs), a category of drugs defined by their use in rheumatoid arthritis to slow down disease progression. Other biologic DMARDs are Enbrel (Etanercept), Remicade (Infliximab), Orencia (Abatacept) and Rituxan (Rituximab). There also are non-biologic DMARDs such as hydroxychloroquine, sulfasalazine, methotrexate and azathioprine, which generally are prescribed to less severe RA patients.

Humira is thought to be the “highest grossing drug in the world. In 2020 alone, AbbVie collected $16 billion in U.S. net revenue for Humira. Today, AbbVie charges approximately $77,000 for a year’s supply of Humira – 470 percent more than when the drug was launched in 2003.”

As a drug that costs $670 or more for a 30-day supply, the CMS has classified Humira as a specialty drug. Humira is not the only specialty drug that is driving higher health costs. According to the CVS Caremark 2020 Drug Trend Report, “specialty drugs accounted for more than half (52 percent) of the pharmacy spend” in 2019. Five therapeutic categories are driving 90 percent of the specialty cost trend for CVS Caremark: oncology, cystic fibrosis, atopic dermatitis, HIV and autoimmune, the category that includes rheumatoid arthritis. Medications are the primary cost driver for RA patients and treating RA is generally expensive due to the cost of Humira and TNF.

GDI Lab Model to Predict Specialty Drug Use

That’s why the Geneia Data Intelligence Lab (GDI Lab) prioritized the creation of models to predict specialty drug use. As we wrote in our abstract to A classification model to predict specialty drug use, “Predicting who is likely to become utilizers of specialty drugs allows care managers to have an early intervention and payers to have financial preparation for the upcoming spending.” It also helps to ensure that the patients who most benefit from taking specialty drugs can access them.

I am particularly enthusiastic about our new work building off these models to predict Humira and TNF inhibitor use in RA patients. Like the Specialty Drug Model, these newer models use administrative claims data where prescription drug use is documented.

In a clinical setting, the number of swollen and tender joint counts, serum levels of C-reactive protein and erythrocyte sedimentation rate, and physical and functional disability often predict poor prognosis in RA patients. In contrast, health plans typically do not have access to this kind of clinical data, and therefore frequently are unable to predict disease progression in RA patients.

Biologic DMARDs such as Humira and Enbrel are advanced therapies for moderate to severe RA patients, and therefore can be used as surrogate for prediction of disease severity and the identification of members for health plan disease management programs. The use of RA biologic medications by members who most benefit from taking specialty drugs can help reduce other medical costs such as hospitalizations. The models can assist care managers to intervene earlier, which improves clinical outcomes and patient satisfaction while reducing unnecessary waste. Because there are a number of biologic DMARDs, health plans and pharmacy benefit managers can use the output from the Humira Prediction Model and the TNF Inhibitor Prediction Model to negotiate value-based care contracts with pharmaceutical companies.

To learn more about the GDI Lab specialty drug use model, visit: https://www.medrxiv.org/content/10.1101/2021.06.30.21259718v1.