A Skeptic's View of "Viagra for Diabetic Women"

OK, this synopsis does not contain much actual information; certainly, not enough to be useful as a basis for medical decisions.  But it does provide an example to illustrate the process of skeptical thinking.  It's from Medscape News (free registration required -- but it's worth the hassle).

Sildenafil May Improve Sexual Functioning in Diabetic Women

NEW YORK (Reuters Health) May 31 - Premenopausal women with type 1 diabetes with sexual dysfunction may find that sildenafil improves arousal, orgasm and sexual enjoyment and decreases pain during intercourse, results of a small pilot study suggest. [...]

To test their theory, they recruited 32 women with type 1 diabetes who in the past had experienced normal sexual desire within their heterosexual relationship, but currently experienced sexual dysfunction -- for 3.5 years on average. [...]

Sildenafil was associated with significantly improved arousal, orgasm, and enjoyment compared with baseline (p < 0.001 for each). Compared with placebo, the experience of arousal (p < 0.01) and orgasm (p < 0.05) were better during active treatment. Only desire and frequency did not change significantly in either group. [...]

Note: sildenafil is the generic name for Viagra.

The original study can be found here, but you need subscription-level access or a bunch of money (that you have no other use for) in order to read the text.

There are a couple of points that one needs to consider in interpreting a study such as this.  First, as a general rule, one should not make medical decisions based upon small studies that are identified as pilot studies.  Doing so, one is likely to expose patients to risks that have not been justified by sufficient research.

Second, it is important to examine critically the outcome measures used in the study.  Specifically, it is necessary to look beyond the measures of statistical significance.  The statistics in this study look reasonably impressive, at first glance.  However, those numbers do not tell you what you really want to know.  Remember, statistical significance does not necessarily translate into clinical significance.  

So let's look at the numbers:

Sexual activity	Baseline	Sildenafil	Placebo	P a	P b	P c
Desire	4 ± 1.3	3.9 ± 0.5	4.1 ± 0.6	NS	NS	NS
Arousal	2.9 ± 1.2	3.7 ± 0.5	3.2 ± 0.3	<.01	NS	<.001
Orgasm	2.8 ± 1.4	3.8 ± 0.8	3.1 ± 0.6	<.05	NS	<.001
Enjoyment	3.5 ± 1.1	4 ± 0.5	3.7 ± 0.6	NS	NS	<.001
Satisfied by frequency	3 ± 1.1	3.2 ± 0.7	3.1 ± 0.4	NS	NS	NS
Frequency of intercourse	1.8 ± 1	2.3 ± 0.5	2.4 ± 0.6	NS	<.05	<.05
Frequency of fantasies	2.1 ± 0.8	2.3 ± 0.4	2.2 ± 0.6	NS	NS	NS
Dyspareunia	2.3 ± 1.1	1.4 ± 0.8	2 ± 0.7	<.05	NS	<.05

• a Sildenafil vs. placebo.
• b Placebo vs. baseline.
• c Sildenafil vs. baseline.

Qualitative items were answered on a 5-point Likert scale, ranging from 1, not at all, to 5, a great deal. Quantitative items were answered as 0, never; 1, less than once a week; 2, once or twice a week; 3, several times a week; 4, once a day/sometimes twice; and 5, several times a day.

To be fair, the authors of the paper are cautious about interpreting the numbers.  They include all the appropriate cautions and limitations.  My concern is that a newspaper could easily pick up this story and print the high points, leading to a great deal of misinterpretation among the general public.  

When I get around to it, I'll put up another post with my interpretation of the numbers.  In the meantime medical students and interested readers are encouraged to look at the data and think about what conclusions may be justified by the numbers.

Depending on when I do get around to it, and when the new Seedlings go live, the follow-up post may be here, or it may be at http://www.scienceblogs.com/corpuscallosum (which is not up as of this writing).