Abstract
Dear sir We would like to respond to the points raised by Lines, Allen and McCarroll in their Letter to the Editor in which they suggest that the time-to-event analysis proposed by Salonen (1) includes only patients who responded. This is inaccurate. As described by Laska and Siegel (2), Salonen's proposed method examines the time to event, be it pain relief or pain-free response, within a given timeframe based on all patients but in a way that provides more information regarding any observed treatment differences. Using a cure model, one assesses the unconditional time to event where patients' observations who never experience the event due to drop out or treatment failure are censored. Also, this cure model allows one to individually assess the two components which have the potential to drive a treatment difference in the unconditional time-to-event analysis—firstly, the probability that the event is experienced during the timeframe studied and secondly, the time to event for those patients who actually achieve the event (conditional time-to-event). Without assessing these individual components, one might conclude that one drug provides an earlier response than the other drug, when this is not the case.
As Salonen explains, in the Goldstein et al. study (3), the only study of the two mentioned by Lines et al. to report an a priori statistical difference (P = 0.046) for the unconditional time-to-event analysis, one may have concluded that rizatriptan had a marginally greater likelihood of earlier relief within 2 h of treatment. When the probability of response, and the conditional time to response are assessed individually, these components demonstrate that the marginal statistical finding was the result of a small difference (6%) in the probability of responding within 2 h, not due to a difference in the likelihood of an earlier response. In addition, Salonen points out that 3% of patients in the rizatriptan group and 1% of patients in the sumatriptan group apparently obtained a first response within 2 h but subsequently lost that response by the 2 h time point. If these patients were considered treatment failures in the unconditional time-to-event analysis, the marginal statistical finding would likely have been lost.
Additionally, in attempts to bolster their story about putative speed differences, Lines et al. report the results of two preference studies which seem to show that patients prefer rizatriptan 10 mg wafer to sumatriptan 50 mg tablets, because of speed of action. However, in this study, speed was the primary reason for preference regardless of whether that preference was for rizatriptan or sumatriptan (4). Further, the actual preference question was biased in favour of the wafer since patients were asked whether they preferred the wafer medication or the tablet medication. As a result the studies are formulation preference studies and are uninformative about differences between the two triptans and thus offer no support to the arguments made by Lines et al.
We also support an evidence-based approach to migraine therapy and believe these comparative trials demonstrate that both treatments are effective treatments for migraine that continue to be under-utilized. The American Migraine II study showed that while 98% of migraine sufferers report that they typically have at least moderate pain with their migraine and 89% report at least some functional impairment, 57% of migraine sufferers rely exclusively on over-the-counter medications for treatment of their migraine.
An evidence-based analysis also shows that treatment is optimized when intervention is migraine specific and given early in the attack. Many patients wait until their attacks have progressed to at least moderate pain before they treat. Burstein et al. (5) have recently suggested medications such at triptans may be most effective when given early in the progression of a migraine attack. Data from retrospective analyses of sumatriptan has shown that early intervention with sumatriptan allows for optimization of therapy to better meet the needs of patients by increasing pain-free efficacy, while reducing recurrence and the need for additional doses of medication (6).
In conclusion, evidence from clinical trials comparing sumatriptan to other currently marketed triptans shows that all have very similar clinical efficacy, surpassing that of nontriptans. Given the similarities in efficacy, factors other than efficacy must be considered in an evidence-based approach to migraine therapy. We believe efforts should be turned away from comparisons of similarly effective drugs and instead toward identifying migraine sufferers who are yet to be diagnosed (52% of sufferers) and toward optimization of treatment in the patients receiving migraine-specific triptan therapy.