Estimated and projected diagnosis rates for dementia in Scotland 2014-2020

A first report into estimated and projected dementia diagnosis rates in Scotland from 2014 to 2020, based on data from health boards in Scotland.

4. Data Sources and Data Collection

Different aspects of Dementia information are captured throughout several datasets; Primary Care, Prescribing, Post-Diagnostic Support, Secondary Care and local systems and therefore a rigorous exercise was conducted first to ascertain the most appropriate and fit for purpose dataset from which to calculate the estimated diagnosis rates. It was concluded that diagnosis of dementia is likely to occur within community and outpatient services, therefore it was presumed the richest source of data would exist within the local systems of each of the Health Boards. Section 4.4 provides more detail on the above datasets.

A number of Health Boards were asked to participate in the project; however, only three of those approached were able to provide the required data items within the specified timescales. These were NHS Ayrshire & Arran, NHS Greater Glasgow & Clyde and NHS Lanarkshire.

An initial analysis of the data found NHS Greater Glasgow & Clyde diagnoses numbers to be grossly inflated in comparison to the two other participating Health Boards - even with consideration to population differences. Further exploration concluded that data submission included records of patients with a dementia diagnosis within community, secondary care inpatient and acute mental health settings, whereas NHS Ayrshire and Arran and NHS Lanarkshire only provided their local data and did not include records from an acute setting (inpatient or mental health) as detailed in Table 2.

Consistency was required across all three sites, therefore it was agreed that ISD would include acute mental health and secondary care inpatient submissions for the other two Health Boards. The reason for this was to ensure that as many people as possible with dementia diagnoses were being captured so as to avoid underestimation of individuals who may require the dementia PDS service.

This alteration meant the new estimates would be calculated from two data sources: local, community-based data and secondary care data. Specific detail on the data collections of each of the data sources is explained in the subsequent sub-sections.

4.1 Local Data Collection

The table below provides details on the data sources from each of the participating Health Boards, including the specific local systems that the data was extracted from as well as any known limitations of the data.

TABLE 2. Local Data Collection Summary Table

Health Board

Time Period of Data Supplied

Local Systems Used

Points of Relevance

NHS Ayrshire & Arran

Apr 2013 - Feb 2015
  • Community Mental Health Team - Elderly ( CMHTE) System

Contains those diagnoses within a community setting. Would only contain inpatient data where a patient was referred from an inpatient setting.

NHS Greater Glasgow & Clyde

1999 - Jan 2015
  • PIMS Outpatient Data
  • SMR04 Data
  • SMR01 Data

Data contained secondary care inpatient and secondary care mental health data as well as community-based data, whereas other participating Health Boards only supplied data from community settings.

NHS Lanarkshire

  • Monthly PDS Submission to ISD

Only 2014 data sent to ISD for the routine PDS submissions was used as the information contained was the same as held locally.

4.2 Secondary Care Data Collection

As a result of secondary care inpatient data being included within the NHS Greater Glasgow & Clyde extract, a decision was made to extract secondary care information for the other two participating NHS Boards also.

Secondary care data is one of the nationally held data sources within ISD.

The secondary care data includes the following:-

  • SMR01 records - General Acute Inpatient and Day Case
  • SMR04 records - Mental Health Inpatient and Day Case
  • SMR50 records - Geriatric Long Stay

As NHS Greater Glasgow & Clyde already supplied SMR01 and SMR04 records along with their local data, SMR50 records were extracted and added to the dataset. For NHS Ayrshire and Arran and NHS Lanarkshire SMR01, SMR04 and SMR50 records were extracted from the centrally held SMR database and added to the local records.

Data were extracted as far back as 1997 where an ICD10 code for Dementia was included in any of the six data items reserved for detailing 'conditions' and where patients resided within one of the three participating Health Boards.

The appropriate International Classification of Diseases (ICD10) codes were supplied by the ISD Terminology Services and are as follows:-

  • F00: Dementia in Alzheimer's Disease
  • F01: Vascular Dementia
  • F02: Dementia in other diseases classified elsewhere
  • F03: Unspecified Dementia

The coding team advised that there are no known quality issues with dementia being falsely recorded on SMR records. However, it is worth highlighting that the existence of an ICD10 code for dementia in one of the diagnosis data items is not necessarily a formal diagnosis of dementia. Dementia is one of the 25 co-morbidities and therefore may be recorded alongside other reasons for admission to acute settings where relevant.

Secondary care data also does not contain a 'diagnosis date' data item, and therefore 'date of discharge' has been used as proxy.

4.3 Summary of Data Collected

The table below summarises high level figures from the data collection process. It indicates the number of individual diagnoses found within each data source for each Health Board. The local Health Board data and extracted secondary care data were combined, and the earliest diagnosis date for each individual was selected utilising CHI number to ensure there was no double counting. Further detail on the linkage and aggregation of the data is explained within Section 5 of this report.

TABLE 3. Summary of Data Collected from each Health Board

Health Board Number of Individual Dementia Diagnoses Identified in Each Data Source in 2014 Estimated Number of People Diagnosed with Dementia in 2014
Local Data Secondary Care Data
NHS Ayrshire & Arran 542 596 1138
NHS Greater Glasgow & Clyde 3362 [2] 27 [3] 3389
NHS Lanarkshire 834 1150 1984

4.4 Other Data Sources Considered

There are a number of data sources available within the Scottish health service held both locally and nationally. In the absence of a specific, official patient level register there was a need to consider the most appropriate options in order to most accurately estimate the incidence of dementia diagnoses. As explained in the preceding sections, the final agreed sources were a combination of localised community-facing data and secondary care data. Other data sources had been considered, but were, at the time, felt to be incapable of providing any significant contributions in comparison with the final data sources that were agreed.

Below is a selection of other considered data sources and reasons for their exclusion at this stage:-

  • Primary Care (eg. SPIRE or QOF) - Primary care data is not currently centrally collated at an individual level and therefore cannot be linked to data from other sources. It would, therefore, be impossible to ascertain whether any identification of dementia in primary care was the first identification, or indeed, when that identification occurred.
  • Prescribing ( PIS) - It was thought that anyone in receipt of prescribed drugs to alleviate symptoms of dementia should have a formal diagnosis of dementia recorded in other data sources, and therefore was deemed inappropriate.
  • Social Care - A formal diagnosis of dementia within a social care setting was felt to be unlikely. As a result focus was directed more toward healthcare data sources.
  • Post Diagnostic Support Data Submissions - The monthly PDS submissions the local Health Boards send to ISD were deemed inappropriate as a measurement of incidence as not everyone who is diagnosed with dementia is contained within these submissions. NHS Lanarkshire's PDS was used for their local data as they confirmed it contained everyone available within their local systems.


Back to top