Combination therapy in the TEMPO study resulted in significantly greater improvements than monotherapy on all measures of PRO: functional status, QoL, and treatment satisfaction. The mean improvement in functional status among combination therapy recipients of 1.0 was significantly larger than in either MTX or etanercept groups. Kosinski et al
suggested that a decrease in HAQ disability index of 0.13–0.24 may be considered clinically significant.18
This is consistent with the 0.22 unit change that Wells et al
determined is perceived by patients with RA as meaningful improvement.19
The difference in HAQ disability index score at 52 weeks between combination therapy and monotherapy is therefore clinically as well as statistically significant. More importantly, by week 52, 58% of patients in the combination group achieved improvement in functional status score of >0.8, and 44% of subjects in the group had HAQ disability scores of
0.5. Further, at week 52, 41% of patients in the combination group were significantly more likely to attain EQ‐5D VAS scores comparable with population norms.
Patient satisfaction with medication is expected to predict drug preferences and adherence to prescription regimens,23
both of which are components of real‐world treatment effectiveness. Patients in the the TEMPO study who received combination therapy or etanercept were significantly more likely than MTX recipients to report satisfaction with their medication.
Response to combination therapy diverged from the response to MTX and etanercept early in treatment and the difference was sustained for 52 weeks. Kaplan‐Meier survival plots for time to occurrence of HAQ scores
0.5 and sustained for 6 months confirmed the faster onset of action in the combination group than with etanercept or MTX alone. Similar trends in improvement over time were found for the EQ‐5D VAS, PGAD, and GHVAS.
Additionally, these results support previous findings that PRO assessments are valid indicators of RA disease status. Table 6 shows that increasing disease activity (measured by the DAS score) is significantly correlated with greater disability as measured by the HAQ disability index, and with lower health status as measured by the EQ‐5D VAS, PGAD, and GHVAS. It has been shown previously that the HAQ disability index correlates significantly with DAS scores.24
The degree of improvement in the QoL with etanercept monotherapy in the TEMPO study is consistent with previous findings. In a 26 week, double blind study comparing etanercept 10 mg or 25 mg against placebo, both etanercept doses produced significantly greater QoL improvement as assessed by the HAQ index and all HAQ subscales except grip.25,26
By 26 weeks, the HAQ index score had decreased (improved) by only 2% in the placebo group, but decreased by 39% in the etanercept 25 mg twice weekly group.25
A similar benefit was experienced by the etanercept group in the TEMPO study, with a 38% decrease from baseline in the HAQ index score at 24 weeks. On a VAS similar to the EQ‐5D VAS, patients in the 26 week study receiving either of the etanercept doses reported greater QoL improvement from baseline than those reported by placebo recipients: approximately 20 points v
8 points, respectively.26
These results are comparable to EQ‐5D VAS improvements in the TEMPO study, in which the mean score in the etanercept group increased by 25 points at 24 weeks and by 26 points at 52 weeks.
The finding in the TEMPO trial of no significant difference in HAQ disability index at 52 weeks between etanercept and MTX agrees with results of a previous study in patients with early RA (diagnosed no more than 3 years before study entry).27
In that study, etanercept yielded a significantly greater improvement in the HAQ eating subscale, while MTX recipients had significantly greater improvement in the dressing subscale; no significant differences were found for other subscales.
The TEMPO study included only patients judged to be appropriate candidates for MTX treatment at study enrolment, so these results may not be generalisable to different patient groups. A potential study limitation is that imputing missing PRO data by the LOCF method may introduce bias if scores change over time.14
However, analysis of the group following the protocol yields results consistent with those for the LOCF group (data not presented).