Ann Intern Med 8 Jan 2007

A systematic review attempts to look at all studies which compare generalist with specialist care. This is doomed for various reasons discussed in the accompanying editorial, not least because most of the studies are observational, very few are randomised, it is generally impossible to adjust for case-mix, and it is very difficult to assess publication bias. Also, this is definitely the sort of review where pears are discussed alongside sea-going mammals. The specialists, of course, almost always get the better press. But do they deserve it?

Here’s a systematic review which deals, thank goodness, entirely with randomised controlled trials, and attempts to find out whether supplements of homocysteine-lowering vitamins improve cognitive function. As readers will be aware, the evidence is disappointing – there is nothing to suggest that B6, B12 or folic acid given to people with normal or impaired cognitive function helps to improve it.

The next systematic review establishes that your C-reactive protein is likely to fall if you lose weight, in case you’re interested.

If someone is dying, who tends to look after them most? Goodness me, it’s their families: who would have thought it? This study even goes so far as to suggest that families often don’t get all the support they need, and that we ought to take that into account in end-of-life care.

Another bit of evidence in favour of influenza vaccine, consistent with other studies: it is associated with a reduction in mortality in adults hospitalised with community-acquired pneumonia.

68 Another contribution to the debate about how patients enrolled in randomised controlled trials differ from real-life patients. We know that in most settings, such as here in the GRACE study of myocardial infarction, the sickest patients tend to have exclusion criteria and so fare worst. Next come the patients who are eligible but not enrolled. In theory these should fare exactly the same as the placebo group of enrolled patients, but in fact they fared significantly worse (OR for mortality 1.6). So there was some factor in favour of enrolled trial patients which these investigators couldn’t identify.

Oddly enough, the first time I heard about thrombolysis was when as a student I heard it mentioned as an experimental treatment for massive pulmonary embolism. More than 30 years later, this paper shows that the question of its effectiveness still hasn’t been settled, at least in the context of submassive PE.