Industry Studies, Publication Bias and Conflicts of Interest
Goldacre cites various reasons why industry should not perform
studies on their own drugs. There are several broad reasons cited, which
more or less boil down to, in his view, intentional bias. He cites a
systematic review of 30 studies which show that industry sponsored studies
are more likely to show favourable results than non-industry sponsored
studies. The reader is invited to assume this means the data is doctored
or some other form of skullduggery is at play.
However there are other explanations, the most obvious of which is
that industry is rather more careful to select drugs which have a good
chance of success. Non-industry sponsors generally do not make the same
level of investment as industry sponsors, and getting it wrong (i.e that
the drug is not efficacious and/or unsafe) is a bigger gamble for
industry. In other spheres the observation would be cause to applaud
industry for its efficiency.
Goldacre should also know that one cause of publication bias
favouring the publication of positive studies is that the journals are
rather less interested in the negative ones. That goes especially for
journals whose former editors most commonly cite publication bias as a
problem. They fail to acknowledge their contribution to that problem.
As far as conflicts of interest are concerned, precisely the same
charge could be levelled at NICE, an organisation whose objective is to
constrain NHS expenditure. Yet few recognise such an obvious bias. I would
expect a study of recent NICE appraisals would show that NICE's estimates
of cost per QALY for new drugs is invariably lower than that of the
sponsors, yet most will blithely assume that the explanation is that
NICE's estimates are correct whereas the sponsors have cooked the economic
data in the same way they cooked the clinical data. But there is an
obvious alternative interpretation, that NICE has an inbuilt bias driven
by its role and underscored by its ideology. Anyone who does not accept
that may be true in at least some NICE appraisals shows just as much bias
as that assumed to apply to industry sponsored studies.
Further, NICE appraisals are not scrutinised with anything
approaching the level of rigour applied to industry sponsored studies by
FDA, EMEA and local country competent authorities.
Competing interests:
Fellow Faculty Pharmaceutical Medicine. Medical Director at a small pharma company
Rapid Response:
Industry Studies, Publication Bias and Conflicts of Interest
Goldacre cites various reasons why industry should not perform
studies on their own drugs. There are several broad reasons cited, which
more or less boil down to, in his view, intentional bias. He cites a
systematic review of 30 studies which show that industry sponsored studies
are more likely to show favourable results than non-industry sponsored
studies. The reader is invited to assume this means the data is doctored
or some other form of skullduggery is at play.
However there are other explanations, the most obvious of which is
that industry is rather more careful to select drugs which have a good
chance of success. Non-industry sponsors generally do not make the same
level of investment as industry sponsors, and getting it wrong (i.e that
the drug is not efficacious and/or unsafe) is a bigger gamble for
industry. In other spheres the observation would be cause to applaud
industry for its efficiency.
Goldacre should also know that one cause of publication bias
favouring the publication of positive studies is that the journals are
rather less interested in the negative ones. That goes especially for
journals whose former editors most commonly cite publication bias as a
problem. They fail to acknowledge their contribution to that problem.
As far as conflicts of interest are concerned, precisely the same
charge could be levelled at NICE, an organisation whose objective is to
constrain NHS expenditure. Yet few recognise such an obvious bias. I would
expect a study of recent NICE appraisals would show that NICE's estimates
of cost per QALY for new drugs is invariably lower than that of the
sponsors, yet most will blithely assume that the explanation is that
NICE's estimates are correct whereas the sponsors have cooked the economic
data in the same way they cooked the clinical data. But there is an
obvious alternative interpretation, that NICE has an inbuilt bias driven
by its role and underscored by its ideology. Anyone who does not accept
that may be true in at least some NICE appraisals shows just as much bias
as that assumed to apply to industry sponsored studies.
Further, NICE appraisals are not scrutinised with anything
approaching the level of rigour applied to industry sponsored studies by
FDA, EMEA and local country competent authorities.
Competing interests:
Fellow Faculty Pharmaceutical Medicine. Medical Director at a small pharma company
Competing interests: No competing interests