Taking anti-depressants may be called 'X' and an increased risk of stroke (from some base rate) may be called 'Y'. Where there is a correlation between an 'X' and a 'Y', there are a number of possible explanations for this correlation other than 'X' causes 'Y'. One is coincidence; another is 'Y' causes 'X' (perhaps those with an increased risk of stroke have a physiological problem that is also liable to make them depressed and so more likely to be taking anti-depressants) and another is that the correlation is explained by a 'Z', which causes both 'X' and 'Y' (imagine 'Z' is a virus that causes physiological damage that at once raises the risk of stroke and the risk of depression). And these basic alternatives may be multiplied because there may be multiple and interacting causes: 'X' and 'Y' may correlate only if A, B, & C hold, and then not if P, Q & R do not hold etc. This indicates why a correlation is only a starting-point for a further causal investigation. 'Double-blind' testing has more to do with reducing experimenter bias than with the problems of working out whether a correlation between an 'X' and a 'Y' is causal or coincidental, and, if causal, whether the causal explanation starts from 'X', or from 'Y', or from some 'Z'. Etc. Being difficult about questions of proof may, however, be motivated less than by a desire for scientific rigour than a desire to protect a vested interest: as in the tobacco industry's attitude to the correlations between smoking and various ailments. Caveat emptor. The trickier question is placing any putative risk in proper perspective, as this requires detailed examination of all relevant data: e.g. if only 1 in a million normally have a stroke, then if 2 in a million on anti-depressants have a stroke, then it could be said taking anti-depressants doubles the chance of having a stroke - but as the risk is still only 1 in every half million it might be thought worth running (because of the increased risks if anti-depressants are not taken) or negligible. The way these things are reported in the press normally raises more questions than are answered, given the paltry information in the report. You may also find your doctor is not of any great help either, although when it comes to the vital question of putting these risks in proper perspective they aught to be (or, according to JLS but not me, aught to burn a letter; if they do, Eric advises they do not post it). Donal London --- On Sun, 14/8/11, Julie Krueger <juliereneb@xxxxxxxxx> wrote: From: Julie Krueger <juliereneb@xxxxxxxxx> Subject: [lit-ideas] Say wha? To: lit-ideas@xxxxxxxxxxxxx Date: Sunday, 14 August, 2011, 19:13 Can somebody parse this for me? <<"Although we found women who took antidepressants were at higher risk, I don't have anything to indicate it's because of the medications," she says.>> http://yourlife.usatoday.com/health/medical/womenshealth/story/2011/08/Depressed-women-have-higher-risk-of-stroke/49931142/1?csp=ylf Julie Krueger