Anti-abortion activists might have failed this summer to reduce Britain's 24-week time limit for terminating pregnancy, but their campaign goes on. In October, the Human Fertilisation and Embryology Bill returns to Parliament and with it will come more proposed amendments to the law.

MPs seeking to restrict access want to require doctors to inform women that abortion can cause psychological damage, and to institute a cooling-off period before the procedure can be performed. They cite studies that purport to show the existence of a “post-abortion syndrome” involving depression and anxiety.

No one would deny that abortion can be deeply traumatic. If there were also a proven link to psychiatric harm, informed consent would certainly require that women be told of it. But as a major research review has made plain, the science does not point that way.

The study from the American Psychological Association (APA), which evaluated all the recent published research, found no evidence that most abortions damage mental health. For one-off abortions in the first trimester of pregnancy there was no elevated psychiatric risk. It was more equivocal about the effects of multiple abortions, but even here it found the case not proven.

This review is not the last word: it was open about the need for further research. But for now at least, “post-abortion syndrome” is at best a hypothesis that has yet to be substantiated. If abortion law is to be based on science, it should not mandate counselling about a putative risk that is not known to exist.

Why, though, did the APA reach this conclusion, given that many studies undeniably support an opposite one? The answer is that much of that research was flawed in design.

Depression that follows abortion is not necessarily depression because of abortion. We all know about the “baby blues”: if there is a higher risk, perhaps pregnancy, not abortion, is responsible. It is thus critical to compare the mental health of women who had terminations with those who completed their pregnancies. Yet not all research has done so.

Poverty, drug use and previous psychiatric illness can be risk factors for both mental health issues and for abortions, especially multiple abortions. Without considering these, it is impossible to tell whether women are becoming depressed because they have had abortions, or because of social problems that lead them to have abortions. Much of the evidence for “post-abortion syndrome” is based on research that lacked such controls.

There is also the question of whether terminated pregnancies are wanted.

There is sound evidence for psychological trauma when women have late abortions because of foetal abnormalities. These, however, are generally wanted pregnancies. They have no bearing on how a woman will feel after an early decision to end a pregnancy she dreaded. But much research has treated wanted and unwanted pregnancies the same way. Poor study design has made it difficult to judge whether or not abortion has an effect on mental health.

This example carries a broader lesson because abortion is not the only issue in which research can deliver a misleading message. All sorts of links that at first look compelling disappear when evidence is analysed closely.

A study might show that people living near incinerators have a higher incidence of cancer. But people with homes in such undesirable locations might also be poor and thus have a higher background risk of disease. Without adequate controls, the research cannot reveal whether their environment is endangering their health.

Such limitations might seem technical, but they are often easy to spot. It is often worth looking for them when trying to figure out whether new research is relevant to personal or public health.

Mark Henderson is Science Editor of The Times

Times recommends

• Jennifer Aniston, time to be a real bitch
• It may be hard to say so, but conversation is back
• News Review interview: Georgina Baillie