Drug companies have been accused of failing to publish drug trials which do not give the "right" result.
Some drug research is not published
Regulatory bodies found it harder to make balanced decisions when negative information was not available, the Lancet medical journal said.
Published research suggested a type of antidepressant drug was safe for children, but unpublished data indicated it was not, a study showed.
The pharmaceutical industry said it was taking steps to solve the problem.
A study by the National Collaborating Centre for Mental Health in London looked at previous research on selective serotonin reuptake inhibitors (SSRIs) for children.
It found that in published studies, all SSRIs appeared to have a favourable ratio of risk to benefit.
But, after also looking at unpublished trials, it was found that, with the exception of fluoxetine, the risks exceeded the benefits.
The Department of Health said last year that most SSRIs should not be given to children.
Tim Kendall, at the centre, said this damaged the role of the National Institute for Clinical Excellence (NICE) in drawing up guidelines.
He said: "Drug sponsors who withhold trial data, or do not make full trial reports available, undermine the guideline programme, which can ultimately lead to recommendations for treatments that are ineffective, cause harm, or both."
The Lancet accused pharmaceutical companies of "confusion, manipulation and institutional failure".
It said Governmental bodies such as NICE require legal powers to ensure biomedical research is used to improve health even if it does not lead to profits for pharmaceutical companies.
The Association of the British Pharmaceutical Industry (ABPI) said it accepted that "things could be better" but that the industry was taking steps to make improvements.
A spokesman said: "We are aware of the problem, we understand that the situation is not entirely satisfactory."
However, part of the blame lay with medical journals which were not keen on publishing negative trial results as they did not make such good news, he said.
The ABPI lists some unpublished trials on its website and a Europe-wide register of all unpublished data will become available to regulatory authorities under a new European Union directive.
The ABPI spokesman added that of the five SSRIs available in the UK, only one is recommended for use in children and the pharmaceutical companies would not be promoting them as this was illegal.
A Department of Health spokeswoman said: "The UK last year advised that the treatment of childhood depression with any SSRI, except Prozac, should not continue, after a thorough review of data.
"The UK is the only country in Europe to have issued comprehensive advice about the use of all SSRIs in children."
She added: "The Medicines and Healthcare products Regulatory Agency supports the need for greater transparency from companies and forthcoming
changes to EU and UK legislation will strengthen the law in this area."
Deborah Glover, editor of the Journal of Wound Care, dismissed the idea that medical journals were reluctant to publish negative trial results.
She told BBC News Online: "I rather think the ABPI spokesman may be confusing the academic, research focused, professional journals with tabloid newspapers.
"I have in previous editorials called for industry to allow the publication of the less than positive results from their research as I believe that not only will this give our readers the information they require, but demonstrates an integrity and confidence on behalf of the industry that facilitates trust between themselves, clinicians and patients.
"The ABPI should now 'name and shame' the journals who appear to be only after the sensational headlines. I would be interested to hear about them as having undertaken a straw poll amongst my colleagues, I can only find those who refute their claims."