This report describes the first stage of a study of the effects of the implementation of crisis resolution teams in England following the policy initiatives first set out in the National Service Framework for mental illness(Department of Health 1998). The work was based entirely on routinely collected data from Hospital Episode Statistics, Department of Health KP90 returns (describing the use of the Mental Health Act) and the annual mapping of English mental health services by the Centre for Public Mental Health at the University of Durham (Glover and Barnes, 2002, 2003, 2005). Key goals for the policy were the reduction of the numbers and length of admissions to mental hospitals and of the use of compulsion. In the work described here, we aim was to establish whether the implementation of crisis resolution teams had been associated with measurable changes of these kinds.
The whole study was restricted to adults of working age (18 to 64). The unit of study adopted for the work on hospital admissions and bed usage was the 303 primary care organisations (PCTs) which between them covered England in 2003. 2002 PCT boundary assignments have been added to Hospital Episode Statistics records dating back as far as 1998/9. These data were thus relatively easily available at this level for the six years from 1998/9 to 2003/4. Compulsory admission data were only available for NHS trusts. Developing time series for trusts over the five years for which these data were available (1998/9 to 2002/3) therefore entailed a complicated exercise tracking the changes in trusts configurations over this period. This produced a data file using 2003 NHS trusts (or in five cases combined ‘super-trusts’) as a unit of study. Service mapping data, indicating the availability and characteristics of crisis resolution and assertive outreach teams could be analysed according to either boundary set, and were available for the years 2001 to 2003.
On inspection all three data sources contained obvious flaws. 19 PCT areas were omitted from analysis because their team provision status was unclear from the service mapping. Hospital Episode Statistics frequently showed single years data missing (or in one instance probably duplicated) for individual trusts. Where there was a clear trend, these defects were filled using calculations based on lines of best fit. In situations where more than one data point was missing or where the trend was unclear (41 cases), PCTs were omitted from analysis. 12 further PCTs were omitted as gender data were largely missing from their Hospital Episode Statistics records. For the main part of the study this left data covering 75% of the population of England.