Study objective-To investigate the rate of suicide in the 12 months after discharge from psychiatric hospital and to determine its relationship to age, diagnosis, and period. Design-Cohort study of patients discharged from psychiatric hospital.

research and local clinical audit before one month survival after acute myocardial infarction can be reliably used for detecting differences in quality of care. In addition, it would be essential to take account of infarct severity. (Heart 1996;76:70-75)

A patient admitted to one Scottish hospital with heart failure may be two to three times more likely to die or be readmitted, both in the short and longer term, compared to a patient admitted to another hospital. Although we may not have accounted for some sources of variation, it is both surprising and disturbing that large, statistically significant, differences in adjusted death and readmission rates can apparently exist for such an important condition in a relatively small country with generally homogenous health care provision. Further, detailed investigation of this apparent variation is required.

An important feature over the last 30 years has been the increasing shortfall in the Conservative vote in Scotland compared with England. The Conservative Party, despite social structural disadvantages in terms of housing tenure and social class, did unusually well until the mid-I950s, particularly among Unionists and Protestants. After considering the historical and religious factors explaining earlier Conservative political strength, it is argued that two factors help to explain the changing politics of the state in Scotland: the establishment of Scotland as a separate unit of economic management in popular perception and the greater dependence on direct state involvement. The Scottish economic dimension has made Scotland an ideological category largely incompatible with Conservative English/British national rhetoric as employed by Mrs Thatcher.The general election of 1987 brought to public attention the growing divergence in the political behaviour of Scotland and England. While the Labour Party could manage only 29.5 per cent of the vote in England, it took 42.4 per cent in Scotland. The Conservatives achieved 46.2 per cent in England, and a mere 24 per cent in Scotland.' This outcome meant that while Labour had 50 MPs north of the border, the Tories could muster only 10. The collapse of the Tory vote in Scotland has appeared to some so natural, has appeared to reflect such obvious truths about Scotland's economic situation and political leanings, that an attempt to account for the Conservative decline appears superfluous. This article argues that there is nothing natural about the Tory collapse. Understanding the divergence in the Conservatives' electoral fortunes north and south of the border requires a careful disentangling of the political, economic and social factors involved. Electoral TrendsFirst of all, let us examine the trends in voting for the two main British parties in Scotland and England at general elections since the war.Two main points can be drawn from the data presented in Tables 1 and 2. First, the most systematic element of electoral divergence between Scotland and England has been the relative decline of the Conservative vote in Scotland. Secondly, the main beneficiary in Scotland of this long-term trend has oscillated between Labour and the Scottish National Party (SNP).

Objective-To determine whether a raised incidence of leukaemia in the Dounreay area occurred in children born to local mothers (birth cohort) or in those who moved to the area after birth (schools cohort) and also whether any cases of cancer have occurred in children born near Dounreay who may have moved elsewhere.Design-Foliow up study. Setting-Dounreay area of Caithness, Scotland. Subjects-4144 children born in the area in the period 1969-88 and 1641 children who attended local schools in the same period but who had been born elsewhere.Main outcome measures-Cancer registration records linked to birth and schools records with computerised probability matching methods.Results-Five cancer registrations were traced from the birth cohort compared with 5 8 expected on the basis of national rates (observed to expected ratio 09, 95% confidence interval 0-3 to 2.0). Ali five cases were of leukaemia (2.3, 0*7 to 5.4). In the schools cohort three cases were found (2.1, 0-4 to 6.2), all of which were of leukaemia (6-7, 1-4 to 19.5). All eight children were resident in the Dounreay area at the time of diagnosis; thus no cases were found in children who were born in or had attended school in the study area but who subsequently moved away.Conclusion -The raised incidence of leukaemia in both the birth and schools cohorts suggests that place of birth is not a more important factor than place of residence in the series of cases of leukaemia observed near Dounreay area.