I have frequently complained in this column about our management’s obsession with assigning me to projects. I would have thought that it was obvious to anybody that my chief value to Pannostrum is as a thinker and detailing me to projects where, inevitably, I have to collaborate with medical advisors, only means that I have to waste my time explaining the bloody obvious to the unenlightened. If at least they kept me on the same project, that would be a small blessing but as it is, they keep on moving me around. In consequence I have to waste time breaking in physicians who have not yet learned to do as the statistician tells them.
So I was not at all pleased to be transferred out of respiratory to diabetes
where I arrived to find the whole division in uproar. On account of various
meta-analyses that have appeared in the medical comics masquerading as
scientific journals, the Agency now requires us to prove that any new
treatment for diabetes does not cause an excess of cardiovascular mortality. In
that connection we were about to publish a meta-analysis of all of our trials
of Sucron® when an article by that well-known
meta-analyst, Percy Vere, appeared in JAM (the Journal
of Appalachian Medicine) and really put the cat amongst the pigeons. An
excess risk of Sucron compared to control was found
using a new approach to meta-analysis and it turns out it was significant
(p=0.04).
I was called to a meeting with the medical advisor for this project, Beatrice Cane, who looked about 15 to me and capable of bursting into tears if I so much as opened my mouth. Luckily, Clive Viper representing Marketing and Di Muddle from Company Regulatory Affairs Pannostrum Pharmaceuticals (CRAPP) were also present, so that there were suitable targets present for my brand of robust enlightenment even if had to tread gingerly for fear of causing any toddlers in the vicinity to bawl.
‘Well,
‘I should explain, Beatrice,’ added Viper, ‘that old
Beatrice looked deeply confused, as if uncertain as to which of Viper and McPearson was the more evil. I think I almost saw her lower lip tremble. I felt sorry for her and decided to be on my best behaviour. The resolution lasted ten seconds.
‘Well, I have reanalysed the data and the results are not significant,’ said Beatrice.
There was a shocked silence. Di turned a whiter shade of pale and even Viper looked as if for two pins he would be gone like a bat out of hell. They both had their eyes fixed on me. Finally Viper said in a sort of strangled whisper, ‘He won’t like that. He doesn’t like it when non-statisticians analyse data.’
‘But I have my own copy of MetOrg and I have done the course and I follow all the instructions in the Archie Association handbook.’
Viper was now looking from Beatrice to me and back again with a sort of horrid fascination.
‘He doesn’t like MetOrg,’ squeaked Di eventually.
‘What’s wrong with MetOrg? It’s very popular.’
‘Count to ten,’ I said to myself, ‘or some self appointed bullying czar will turn up on television claiming Pannostrum has a bullying culture and “is in denial”.’
I put on my most ingratiating smile. Viper looked as if he
was going to be sick. ‘What was the best-selling hit of 1969, I said?’ Viper
relaxed. ‘
‘Ill give you a clue’, I said, ‘The group had a highly apposite name. They were called The Archies.’
‘Oh,’ said Di, ‘I remember that. It was jolly good. Sugar, Sugar.’
‘Yes, Indeed’ I said, ‘Sugar Sugar. However, I don’t know about good. It was a catchy number and certainly popular. However, although it outsold it by far, it hardly ranks with an earlier number one that year. I refer, of course, to that classic of Credence Clearwater Revival’s, Bad Moon Rising which has frequently formed an attractive feature,’ I added modestly, ‘of karaoke night at annual conference. In other words, popular is not the same as good.’
Beatrice looked nettled. ‘Well that still doesn’t explain why I am getting different results to Percy Vere. I have looked at my MetOrg analysis and it seems very similar to the company report.
‘Yes. It’s all very disappointing,’ I said.
‘What?’ Said Viper.
‘Well two things are very disappointing. The first is that JAM should publish a method that is so silly. I have not seen anything quite as barmy since the Agency’s obsession with type III sums of squares back in the 1980s. It can have very paradoxical features,’ I continued, warming to my task. ‘Suppose you had a situation where a number of very large trials had been run and you had performed a meta-analysis with a tight confidence interval. Now some tin-pot investigator-sponsored trial is run with hardly any patients, you have to add it to the mix and the expected width of the confidence interval; will actually rise. It must be one of the few cases in scientific investigation where more information is worse than less. So yes, you are right, if you analyse the trials properly there is no significant excess risk.’
‘So the analysis was disappointing, that’s one thing, said Viper, ‘but you mentioned two. What’s the other?’
‘The correct result’, I said, ‘It’s not significant.’
‘But that’s good. What could be wrong with that?’
‘Well consider this. Diabetes is a risk factor for cardiovascular mortality and Sucron is an effective treatment for diabetes so don’t you think that something is rather disappointing?’
‘What?’ said Viper and Muddle together.
‘Well, call me old-fashioned but shouldn’t it actually save lives?’
DISCLAIMER This is a work of fiction. Any resemblance to event or persons is pure coincidence.