results. There was, however, greate

results. There was, however, greater loss to follow up in the TTM
arms than the control, as is common in these studies, which is
probably due to the greater intensity of the TTM arms. Sensitivity
analysis for loss to follow up changed the results somewhat
because of the assumptions about those with missing data.
However, while it changed the overall estimates of effect, it did
not change the relative effectiveness of the TTM to control overall
or across baseline stage, so bias from this is an unlikely cause of
the null results.
These results are consistent with others that found little effect
of TTM-based interventions, but there is discrepancy across trials.
The positive results have largely come from Prochaska et al. (e.g.
Prochaska et al., 1993, 2001; Velicer et al., 1999), while the
negative results have been produced by other researchers (e.g.
(Aveyard et al., 2003; Lawrence et al., 2003; Lennox et al., 2001)),
with most results suggesting only small to moderate effects from
stage-based interventions relative to control (Riemsma et al.,
2003). We used identical interventions to Prochaska et al., so the
discrepancy cannot arise from variation in intervention effects.
The effects of the TTM-based intervention relative to control by
stage of change are consistent with all previous analogous findings
(Aveyard et al., 2006; Prochaska et al., 1993). These two
studies found evidence that the effects of stage-based interventions
arise primarily because those who are ready to change
respond to the stage-based interventions better than those who
are less prepared to do so, which is contrary to the hypothesis
advanced that TTM-based interventions work by effectively
assisting those not ready to change behaviour (Prochaska &
Velicer, 1997). In this study, this pattern was less apparent. In
these three trials, the stage-based intervention was more intensive
than the control intervention, with more contacts and more
literature. A parsimonious conclusion then is not that stage
matching to those that are not ready to change is important;
rather, that more intensive interventions given to those who are
intending to change are more effective.
In conclusion, these results show that one TTM-based intervention
was not more effective at inducing stage movement, and
that, contrary to the TTM predictions, was not more effective in
those in the early stages of change. Analysis of the content of these
interventions and the findings that are counter to the theory
suggest that the mechanism of action of the TTM-based interventions
needs to be re-thought.
Acknowledgments
The study was funded by the health authorities of the West
Midlands. Paul Aveyard is funded by the National Institute of Health
Research.