A response – published in the British Medical Journal – to the argument put forward by Jauhar et al who support the current scale of antidepressant prescribing, including for mild conditions, and for those provoked by social and economic conditions:
We thank Jauhar and colleagues for their comments on our letter. Their overall point seems to be that the current scale of antidepressant use is not of concern, and that the use of antidepressants for mild conditions and those provoked by social and economic conditions is justified. We address their points in order.
Antidepressants lead to a range of adverse events as evidenced in matched comparisons of antidepressant users showing a higher incidence of increased weight gain and increases in falls, cardiovascular disease, bleeding and mortality. These studies carry a risk of residual confounding but they are consistent and highlight the lack of long-term randomised trials that can establish the incidence, prevalence and severity of antidepressant adverse effects definitively, which is a concern for all, but particularly those with milder conditions. A precautionary approach would be prudent. More than 50% of patients using common antidepressants experience treatment-emergent sexual dysfunction in double-blind randomised controlled trials,3 which, according to the EMA, can persist after cessation.
They question the extent of withdrawal symptoms. They give a “worst case” scenario figure of symptoms in 25% of people. This still represents millions of people but the evidence they cite is for patients who have been on antidepressants for just 17.2 weeks on average. The median duration of antidepressant use is more than 2 years in the UK  and there is a clear dose-response relationship between duration of use and risk of withdrawal effects, with incidence and severity rising after 6 months of use. Double-blind randomised controlled trials (conducted by drug companies) demonstrate that 50% of patients stopping antidepressants experience withdrawal effects. The RCT in the NEJM does not measure the severity of withdrawal per se, but usefully demonstrated withdrawal effects that lasted for months on average, consistent with a protracted withdrawal syndrome. …”
You can read more from here.