My first reaction to this week's front-page headline, “Depression drugs don't work”, was surprise. I took Prozac only once, but it was transforming. I know several people who have found the same. One believes that it saved her from suicide. Her depression has been, at times, a crippling illness.

My second reaction was alarm. No sooner had the new study come out than it was being presented as a morality tale of corrupt pharmaceutical companies marketing ineffective drugs to a population of the worried well who should bloody well be pulling their socks up. The “no pain, no gain” lobby were out in force, claiming that depression serves a useful purpose, because it tells us what we need to hear if we are to fix our lives. I wonder if they've ever had to live with a sufferer. Real depression is a state you can't just drag yourself out of. Clearly, drugs work only for some people. But should we deny anyone the chance to see if they are among the lucky ones?

I read the University of Hull study. It analyses the clinical trials of six new generation anti-depressants called SSRIs (selective serotonin reuptake inhibitors), carried out before those drugs were licensed. Using unpublished data that the manufacturers had sat on, the researchers concluded that these drugs worked no better than a placebo for the majority of patients with moderate depression, and only marginally better for those who were severely depressed. It is damning stuff.

It doesn't mean that no one got better. Many did. But the control group did almost as well on placebos, which duplicated “more than 80 per cent of the improvement seen in the drug groups”. Even among the most severely depressed, who were less responsive to placebos, the improvement in the control group was still “significant”. This compares, the authors say, with the placebo effect on pain which “is estimated to be about 50 per cent of the response to pain medication”. Wow. If placebos are so good at relieving pain, and even better at relieving depression, why aren't we studying placebos?

There is clearly a great deal we still don't know about the brain. So while it is disgraceful that the authors had to fight pharmaceutical companies to get unpublished and unfavourable data, I am still not sure that the industry is defrauding the public by pushing happy pills. As the Harvard psychiatrist David Healy pointed out in Let Them Eat Prozac, different personalities react to different pills in vastly different ways. Some people spring back to life and others slump. The problem with Hull's meta-analysis is that it obscures individual variations by lumping all the data together. It doesn't tell us that no drugs work for anyone, but nor does it shed much light on what works for who.

I found a US study called Star*D, funded by the respected National Institute of Mental Health, which has followed 4,000 patients for seven years. Thirty per cent of these recovered on the first drug and another 21 per cent after changing drug. A 50 per cent cure rate looked good to me. But that research has been attacked for not including placebos, which some psychiatrists say would have had a similar effect. According to a Canadian task force, “spontaneous remission can occur over six to twelve months in up to 50 per cent of affected people”.

So what are we to believe? On the one hand we have people clearly crippled by depression. On the other we have people bouncing back to health with no help. There is clearly a confusion between depression and sadness, a normal human response to difficulties that doesn't need to be medicalised. Every doctor I have talked to thinks that medics have become slap-happy about happy pills. They turned from Valium to SSRIs because they were less addictive. But should we go as far as Irving Kirsch, from Hull, when he asserts that “there seems little reason to prescribe antidepressant medication to any but the most severely depressed patients”, when we still seem to know so little about how these pills work and on who?

Sixteen million prescriptions for antidepressants were written in the UK last year, for 3.5 million people. That's a lot. But 38 million prescriptions were written for antibiotics, of which a higher proportion were probably unnecessary. And antibiotics have terrible side-effects. They have bred resistant bacteria that are killing people in hospitals. Yet I haven't heard anyone argue that a good dose of flu makes you a better person, or that we should restrict antibiotics only to people with pneumonia.

It is the eagerness to stigmatise antidepressants that makes me wary. It is easy to forget how revolutionary these drugs have been. In the 1960s the first generation of antidepressants released hundreds of thousands of people from the hell of mental hospitals. The idea that depression was a chemical imbalance in the brain that could be treated was a liberation. Prozac becoming a household name has been an enormously powerful way to reduce the stigma of mental illness.

Having read the research, I suppose my own boost from Prozac was a placebo effect. If so, I guess it'll never work for me again. But bravo the placebo, while it lasted. I didn't need to wait two years for therapy on the NHS: I needed a quick fix to regain equilibrium. I dislike our tendency to moralise about people who occasionally want a short cut to deal with reality. Britain's traditional approach to self-medication is alcohol: I have seen the effects of that, and I would take the happy pills over that any day. Even if their benefits are all in the mind.

Times Recommends

• It’s not just about Iraq, it’s about Blair’s style
• You can’t rank hospitals like football teams
• Minarets are not an essential part of Islam