The early days of behaviour therapy were characterized by a striving for scientific respectability, mainly by carrying out controlled evaluation of treatment using group designs.
For much of my earlier career I was involved in such treatment outcome research. At that stage I never doubted that it was the right thing to do. However, subsequent and somewhat painful salutary experiences led me to attach over-riding importance to the marked prognostic variability in couples presenting for help with sexual problems, and the need to control for such variability in any attempt at a methodologically sound outcome study.
When in Oxford I was involved in two attempts at controlled outcome studies. In a randomized treatment study of 36 couples with sexual problems, three treatment methods were compared: The second attempt was possibly the first study to assess the combination of pharmacological and psychological treatment, comparing two pharmacological treatments in the process, testosterone T and diazepam Carney et al No differences in outcome were found between the weekly and monthly sex therapy groups.
However, women receiving T showed significantly more improvement than those in the diazepam group. This left us with the impression that oral T was an effective treatment at least in combination with sex therapy. However, two attempts to replicate this finding failed. In each case T was compared not with diazepam but with placebo. One possible explanation for this failure to replicate was that in the original study diazepam was having an adverse effect, possibly by reducing the effectiveness of counselling, whereas T was having no effect.
In Edinburgh we carried out a comparable study with male sexual dysfunction Bancroft et al , comparing T with placebo, both in combination with sex therapy. We found trends in favour of placebo; but there is no reason why T should be less beneficial than placebo in men.
We also found that in spite of randomization there were differences in the two treatment groups i. T and placebo before treatment in variables that were found to have prognostic significance. In some cases these pre-treatment differences reached statistical significance. It was apparent that they could have accounted for the post-treatment differences, making the results uninterpretable.
We had been unlucky in our randomization. But this experience emphasized the importance of balancing for pre-treatment variables of prognostic significance. In our case we had an imbalance of statistically significant proportions. But such significance levels are arbitrary and their biasing effect on the results depend on the degree of the difference, not on whether it is statistically significant.
If prognostic variability is not controlled, there are two important negative consequences. If one treatment group has a better prognosis than the other, bias will be introduced, inflating or cancelling out real treatment effects or producing spurious effects where none exist as probably happened in our ill-fated study. Secondly, if prognostic variability is not taken into account in the analysis it will add to the experimental error.
Since the statistical significance of a treatment effect is measured in relation to this residual error, moderate treatment effects will remain undetected, unless the sample size is increased substantially. Anyone who has clinical experience of sex therapy will know how variable couples are in their response to treatment.
Some improve with apparent ease and little therapeutic effort; others are notably resistant to change. Thus, it is not difficult for genuine and worthwhile treatment effects to be obscured by the prognostic variability.
This experience confronted us with the need to first identify and then control for the key prognostic variables. There have been a few, more recent controlled trials of cognitive behaviour therapy CBT for sexual problems.
Van Lankveld et al randomly assigned women with primary vaginismus to cognitive-behavioural group therapy, cognitive-behavioural bibliotherapy or a waiting list. Twenty-one per cent dropped out before completing treatment.
The authors concluded that there was a modest treatment effect, and commented that it was much lower than that reported in uncontrolled studies. They suggested that the 3-month duration of the treatment programmes may have been too short, and that their cases may have been more severe than other case series. Apart from no consideration of potential prognostic factors, I have two principal concerns with this study. A waiting list control seems inappropriate when assessing the impact of treatment on a life-long condition of this kind, in contrast to medical or psychological conditions of more recent onset that have a greater likelihood of spontaneous remission.
More important are the limitations of the treatment methods involved. Clearly, they were informative for the couples, and this is an important component. In addition, they involved a hierarchical approach to vaginal penetration, with the woman's finger, subsequently her partner's finger and a plastic dilator preceding attempts at penile insertion.
However, a potentially important difference when compared with the treatment described on p. This, if done sensitively, could make a crucial difference to reassuring the woman and facilitating her homework assignments of this kind, though it does restrict this component of treatment to a therapist experienced at vaginal examinations.
Further controlled studies of vaginismus should control for this potentially crucial component. Couples were recruited through newspapers and 74 couples participated. A reasonably precise set of criteria for the diagnosis of HSD was used, which allows comparison with subsequent studies. Thirteen established questionnaires were used to assess mood, marital, psychological and sexual functioning.
The treatment involved 12 weekly group couple sessions 4—6 couples per group and a pair of male and female therapists. The control group was a waiting list; half of the couples were assigned to wait 3 months and were assessed at the beginning and end of the wait period before proceeding with the same treatment.
However, it is not clear how many of the 74 women were assessed at that stage; only 45 of them completed all the questionnaires. This is a useful study, with a number of strengths, including detailed possibly too detailed assessment and reporting of couples' reactions to the various components of the treatment. The method of recruitment may be important, and one wonders how such a group would compare with those referred to a clinic for treatment. The value of delaying treatment for 3 months as a form of control is limited, and only the one treatment programme was used.
Once again, no information is given about the characteristics of those who did and did not respond, which could be of prognostic relevance. A controlled comparison of three methods of treating vulvo-vestibulitis Bergeron et al , is considered in Chapter 13 see p.