Definition
Percentage of patients surviving for 30 days after emergency admission with principal diagnosis of stroke.

Aim
To provide an overview of risk to patients diagnosed with stroke.  It should serve as a useful starting point to highlight issues or raise questions about stroke survival, which might merit further investigation.

Period of coverage
Data are presented for patients discharged in each of the years ending 30th June 1999 to 30th June 2008.  

Data sources
This indicator is based on the linked data set of SMR01 (Scottish Morbidity Records) and death records held at ISD Scotland (Kendrick and Clarke, 1993).  SMR01 is a dataset for general acute inpatient/day cases. The linkage and use of the death records is by permission of the General Register Office for Scotland (GROS).

Deprivation data are used in the calculation of standardised survival rates (see below).  For Scottish Clinical Indicators on the web, published from January 2005 onwards, deprivation quintiles are based on the Scottish Index of Multiple Deprivation (SIMD), derived from the 2001 census data.  Full information on the SIMD can be found on the Scottish Executive website; report here and the dataset here.
 

Criteria for inclusion 
The indicator is for patients admitted as an emergency with a principal diagnosis of stroke.  The International Classification of Diseases and Related Health Problems Tenth Revision (ICD-10) codes taken as indicating stroke are as follows:
 

Emergency admissions
From the July 2009 update the definition for selecting emergency admissions in SMR01 data has been revised. The previous definition selected old type of admission recorded as emergency. The new definition can be found here. There may be changes to figures if comparing this update to previous updates.

Patient basis of the indicator
Each patient admitted with a stroke is counted only once within each year no matter how many times the patient is admitted for stroke within that period.  The aim is to avoid any double counting of the same patient.
  

Definition of outcome
Outcome is defined as survival for at least 30 days after admission.  Mortality is established on the basis of linkage to the GROS death records and so takes account of any deaths occurring after discharge from hospital. 
 

Standardisation
Results are indirectly standardised for age, sex and deprivation (SIMD) quintile.  The reference population in terms of which the rates are standardised is the total number of strokes included for Scotland over the entire period.  The rates are thus standardised over time as well as between Hospitals/NHS Boards.  For a more detailed explanation of the indirect standardisation process, please refer to annex 7 of the May 2002 Clinical Outcome Indicators Report (Clinical Outcomes Working Group, 2002)
 

Presentation at NHS Board and hospital level
The indicators are presented at NHS Board and hospital level. Data for NHS Argyll & Clyde totals are presented separately for NHS Argyll & Clyde (H) and NHS Argyll & Clyde (GG). All hospitals are included in NHS Board and Scotland totals.  However, data for individual NHS Boards and hospitals are presented only if the total number of strokes included in the indicator was greater than 50 in the year ending 30th June 2007 or the year ending 30th June 2008.  The threshold for inclusion was chosen partly for confidentiality reasons and partly to show most of the major organisations within acute care.

Figures at NHS Board and hospital level are available here as a series of downloadable excel.

Inclusion of data for Queen Margaret Hospital
Please note that Queen Margaret Hospital data has been included in the downloadable file for this indicator, even though it was excluded from the Clinical Outcome Indicators report of May 2002 (Clinical Outcomes Working group, 2002).  The reason for the exclusion in the May 2002 report was that there were doubts relating to the coding of strokes in the years prior to 2002.  These doubts are still present for the first two years of data presented (1998 and 1999).
 

Interpreting the figures: general points
Interpretation of indicators presented by single year should be carried out with caution.  Statistically hard and fast rules cannot be provided for such interpretation - it is a matter of common sense. In particular, very little should be read into data for a single year, especially when this consists of a deviation from an established pattern.

The SMR01 source data is 99% complete for year ending June 2008 at Scotland level; however further data may still be submitted by some hospitals.  Therefore, any apparent fall in the figures for 2008 as compared with previous years, must be viewed with caution. SMR data completeness by NHS Board can be found at here.

Deaths records from the GRO are final up to December 2007 and 2008 data are provisional. Final data from the GRO does not tend to vary a great deal from provisional data.

Confidence intervals for the data for individual years have not been presented.  This is not only to keep the presentation simple but also to avoid looking at individual years data and instead focus attention on the long-term trends.  The Clinical Outcome Indicators Report of July 1999 (p20-27) contains a detailed discussion of the kinds of inference it is legitimate to draw from patterns of apparent change and stability in the indicators.
 

In terms of the annual trends presented here, the larger, longer lasting and more stable the difference between the indicator for an NHS Board and the Scottish mean, the more likely it is that this represents a real difference in outcome rather than a chance difference reflecting random variation over time.

Most importantly, however, even when a large, stable and long-lasting difference in outcome is apparent, the caveat, which has been stressed throughout the publication of such survival indicators, applies equally here.  No direct inferences about quality of care should be drawn from the data presented here.  However, it should serve as a useful starting point to highlight issues or raise questions about outcomes in this area which might merit further investigation. The effects of any differences between hospitals in aspects of case mix for which we have been unable to standardise, such as mean severity on admission or differing diagnostic thresholds, may be just as significant as any differences in quality of care.

Interpreting the figures: specific points
Previous Outcome Indicators Reports (pdf versions are available here) have discussed at length the necessity of recognising the extent to which apparent differences in outcome may be due to differences between hospitals in the case mix of patients admitted over and above those aspects of case mix for which standardisation is carried out.  Such additional case mix factors may reflect differing admission thresholds, differing diagnostic procedures or differing coding conventions.

In particular, the 1999 Report discussed at length the possible reasons for the wide discrepancy in survival rates after admission for stroke between the Royal Infirmary of Edinburgh (RIE) and the Western General in Edinburgh.  It was pointed out that a form of patient selection might have been partly responsible.  The closure of the Accident and Emergency Unit at the Western General meant that more severe emergency cases of stroke were more likely to be taken to the RIE.  Since the publication of the 1999 Report another possible factor has been suggested which may help explain the apparently low survival after stroke at the RIE in the mid-1990s.  It may be that in some years in the mid-1990s, milder stroke cases may not have had their stroke diagnosis recorded on the SMR01 record for the first episode of their stay in hospital (the indicator is defined in terms of principal diagnosis on the first record of the stay).  This may have been because of the introduction of the Acute Medical Admissions Unit at the RIE involving a relatively short episode of care as the first episode of the stay. This short initial episode may have created local difficulties in getting a definitive diagnosis onto the relevant SMR01 record. 
 

References
Clinical Outcomes Working Group (1999)   Clinical Outcomes Indicators July 1999. Edinburgh: The Scottish Executive.
 
Clinical Outcomes Working Group (2002) Clinical Outcomes Indicators May 2002. Edinburgh: The Scottish Executive.