Background/Purpose:

The Outcome Measures in
Rheumatology (OMERACT) Initiative has suggested an analysis of the occurrence
of benefit and harm in trials simultaneously, at the individual patient level can
enhance trial reports. We applied this method in a recent trial dataset.

Methods:

OMERACT suggests creating 3
levels of benefit and harm, and expressing each patient’s outcome as a pair of
values for benefit and harm. Results for the groups are summarized in 3×3
contingency tables. In addition, outcome of patients at the extremes can be
labeled as ‘unqualified success’, resp. ‘unmitigated failure’. The 48-week data
of the recent ‘RA Comparison of Active Therapies’ (RACAT) study were
re-analyzed according to this framework.

Three levels of harm were defined
as 1) no adverse events (none), 2) at least one adverse event (AE), and 3) at
least one serious adverse event or withdrawal from the trial due to an adverse
event (SAE).  The three levels of
benefit were defined using both ACR and the EULAR response criteria.  Under ACR criteria, the three levels of
benefit were defined as 1) less than 20% improvement (none), 2) 20% or 50%
improvement (moderate), and 3) 70% improvement or greater (good). Unqualified
success was defined as achieving a good response without AE;
unmitigated failure as no response with SAE. Chi-square tests assessed the
association between benefit and harm, or exact tests where appropriate.

Results:

The trial compared addition of
sulfasalazine and hydroxychloroquine or etanercept to methotrexate (MTX) in RA
patients with inadequate response. Of the 353 randomized patients, 309 were
followed to study completion at 48 weeks and were eligible for analysis.  Of those, 226 (73%) experienced adverse
events and 46 (15%) experienced serious adverse events.

Under the ACR criteria, 3%
patients experienced an unqualified success, while 9% experienced an
unmitigated failure.  For the 154
patients randomized to triple therapy, the rates were 3 resp. 9% (Table); for the 155 MTX + etanercept the rates were 3 resp. 8%
(Table).

Under EULAR criteria, 6% patients
experienced an unqualified success while 4% experienced an unmitigated failure.
For the triple therapy patients the rates were 4 resp 4%;
for the MTX + etanercept patients the rates were 6 resp. 4%.

Benefit by both EULAR and ACR
response distributions were significantly associated with the three levels of
harm (p=0.008 and p=0.01, respectively): frequency of AE and SAE increased as
response decreased.  There we no
significant differences in associations across treatment groups (p=0.83 and
p=0.44 by EULAR and ACR criteria, respectively).

Conclusion:

This is the first study to
suggest that in trials the occurrence of harm increases as benefit decreases.