Skip navigation

Mindstretcher - trials or systematic reviews? [Oct 1998; 56-10]

From time-to-time questions are asked about why meta-analyses and randomised trials do not always agree. Bandolier found itself scratching its head about why this seemed to be the case regarding two recent publications, one a meta-analysis [ 1 ] which concluded that nonoxynol-9 prevented sexually transmitted infections, and a randomised trial [2] which concluded the opposite.


Nonoxynol-9 is a nonionic detergent that works as a spermicide by disrupting the membranes of epithelial and sperm cells. It has been seen to do much the same to sexually transmitted pathogens in the test tube, including viruses like Herpes simplex and HIV.


The meta-analysis [1] looked for all trials and cohort, case-control or cross-sectional information. Included studies had to have original data to allow calculation of relative risks and spermicides containing nonoxynol-9 had to be used separately from other barrier contraceptive methods.

Twelve articles were identified. For six randomised trials looking at gonorrhoea (1881 women) the relative risk of transmission was 0.62 (95% confidence interval 0.49 to 0.78). Four randomised trials evaluating Chlamydial infection had a relative risk of 0.75 (0.62 to 0.91). There was insufficient information for HIV.

So clearly nonoxynol-9 spermicides reduced transmission of sexually transmitted infections by 30-40% in randomised trials (confirmed by observational studies). Almost all were conducted among sex workers or health clinics with high rates of sexually transmitted infection.

Randomised trial

This was conducted in female sex workers in Cameroon. It was beautifully designed and conducted, and women (1170) were randomised to use of nonoxynol-9 film plus male condoms or placebo film plus condom for 12 months with monthly check ups and data collection. DNA probes were used to test for sexually transmitted infections.

At the end of the one-year trial there was no difference in infection rates for gonorrhoea, Chlamydia or HIV.

Spot the difference?

So why the big difference, and which was right? The eagle-eyed of you will have seen that the meta-analysis excluded barrier contraceptives (that is, it was testing the theory that nonoxynol-9 reduced sexually transmitted infections). The randomised trial positively encouraged (and monitored) use of condoms.

Chalk and cheese

There probably isn't any real controversy here. It's just that the meta-analysis and the randomised trial were testing different hypotheses in different circumstances. Why should the results be the same?


  1. RL Cook, MJ Rosenberg. Do spermicides containing nonoxynol-9 prevent sexually transmitted infections: A meta-analysis. Sexually Transmitted Diseases 1998 25: 144-50.
  2. RE Roddy, L Zekeng, KA Ryan et al. A controlled trial of nonoxynol 9 film to reduce male-to-female transmission of sexually transmitted diseases. New England Journal of Medicine 1998 339: 504-10.

previous story in this issue