Duration of DMARD and Biologic Therapies:

1
Multivariate Predictors of H. Zoster
Duration of DMARD and Biologic Therapies
Switching Behavior and Predictors of
Discontinuation Frederick Wolfe and Kaleb Michaud
National Data Bank for Rheumatic Diseases,
Wichita, KS
Abstract PURPOSE. The duration of therapy is a
useful guide to the comparative, real life
effectiveness of that therapy. Therapy is
continued when the real or perceived value of the
therapy is at least as great as the real or
perceived value of alternative therapies.
However, other factors influence treatment
durability, including cost, insurance status,
severity of illness and previous therapy. We
included these factors in analyses of new
therapies to provide an additional view into
their use and effectiveness. METHODS. Patients
in a long-term study of RA outcomes who were
starting new treatments (etanercept (N1,716),
infliximab (N1,757) and leflunomide (N1,758)
were assessed semiannually. Not enough patients
used adalimumab for analysis. Time to treatment
discontinuation was determined by survival
analysis. Predictors of discontinuation were
analyzed using Cox proportional hazards
regression using baseline and lagged time-varying
predictors. RESULTS. The median time to initial
treatment discontinuation was longest for
etanercept (4 years), followed by infliximab (3.5
years) and leflunomide (2 years) However, some
patients resumed therapy after a break in therapy
of 6 or months or more, and this would have
increased median time of therapy by approximately
6 months. Of 527 patients discontinuing
etanercept, 49 started on adalimumab and 145
began infliximab. Among 569 patients
discontinuing infliximab, 47 used adalimumab and
61 used etanercept. Of 866 patients continuing
leflunomide, 102 used infliximab and 163 used
etanercept. Regardless of which treatment was
terminated, there was little change in
methotrexate or prednisone use. Patients
beginning infliximab had more severe RA at
treatment onset than patients receiving
etanercept. In addition more infliximab patients
were gt65 years of age and receiving Medicare
(45.8 vs. 22.1), had lower median household
income (35,000 vs. 45,000) and had slightly
less education (13.4 vs. 13.7 years). Values for
these variables for persons starting on
leflunomide were 33.6, 35,000 and 13.5 years.
The rate of discontinuation was related to
initial RA severity, previous DMARD/biologic use,
prednisone use, comorbidity and the presence of
concomitant fibromyalgia. However,
discontinuation was even more strongly related to
severity measurements over time, indicating that
failure to respond acceptably was a major factor
in discontinuation. Prednisone use, HAQ score and
a composite disease activity index were the
strongest predictors of time influenced
discontinuation. CONCLUSIONS. The median time to
initial treatment discontinuation was 3.5 to 4
years for biologics and 2 years for leflunomide.
A minority of patients discontinuing biologic
therapy switch to another biologic. However,
switching was more common with etanercept than
infliximab. Baseline and ongoing RA severity
predicts discontinuation, and differences in
discontinuations between infliximab and
etanercept reflect baseline severity and
socioeconomic differences.
Characteristics of Patients at Treatment
Initiation
Time-Varying Predictors of Discontinuation

Patients beginning infliximab had more severe RA
at treatment onset than patients receiving
etanercept. In addition more infliximab patients
were gt65 years of age and receiving Medicare
(45.8 vs. 22.1), had lower median household
income (35,000 vs. 45,000) and had slightly
less education (13.4 vs. 13.7 years). Values for
these variables for persons starting on
leflunomide were 33.6, 35,000 and 13.5 years.
Rates of Treatment Discontinuation


The rate of discontinuation was related to
initial RA severity, previous DMARD/biologic use,
prednisone use, comorbidity and the presence of
concomitant fibromyalgia. However,
discontinuation was even more strongly related to
severity measurements over time, indicating that
failure to respond acceptably was a major factor
in discontinuation. Prednisone use, HAQ score and
a composite disease activity index were the
strongest predictors of time influenced
discontinuation.
Treatment Switching Following Discontinuation

All D/C refers to patients who restarted therapy
after initial discontinuation. 25 and 50 is the
time in years until 25 or 50 of patients would
have discontinued treatment, Adalimumab is not
included in these analyses because of
insufficient numbers owing to more availability
of that treatment.
CONCLUSIONS. The median time to initial treatment
discontinuation was 3.5 to 4 years for biologics
and 2 years for leflunomide. A minority of
patients discontinuing biologic therapy switch to
another biologic. However, switching was more
common with etanercept than infliximab. Baseline
and ongoing RA severity predicts discontinuation,
and differences in discontinuations between
infliximab and etanercept reflect baseline
severity and socioeconomic differences
The median time to initial treatment
discontinuation was longest for etanercept (4
years), followed by infliximab (3.5 years) and
leflunomide (2 years) However, some patients
resumed therapy after a break in therapy of 6 or
months or more, and this would have increased
median time of therapy by approximately 6 months.
National Data Bank for Rheumatic Diseases