Data Quality

Strengths and limitations

In order to determine the quality of the statistics that we publish, PHS assesses the risk of data quality concerns (external website) for each publication. Publications are assigned a low, medium or high data quality risk rating. This rating is based on factors such as the number of data suppliers involved, the complexity of the data collection process and the quality assurance checks applied to the data. This publication has been assigned a low risk of data quality concerns.

How can the data in this publication be used?

This data can be used to:

  • Compare areas across Scotland
  • View trends over time
  • Compare activity between different specialties, age and sex groups, or deprivation levels
  • Assess whether patients were treated within or outwith their own Health Board
  • Explore the relationship between hospital activity and living in a deprived area

This data should not be used to:

  • Directly compare with other UK countries
  • Identify areas of affluence
  • Identify how much more deprived an area is from another

Data collection and validation

The data presented in this publication are sourced from hospital administrative systems across Scotland and extracted from the Scottish Morbidity Record (SMR) 01 and 04 datasets held by PHS (formerly ISD Scotland).

The Scottish Government target for SMR submission to PHS is 6 weeks following a hospital admission or discharge, or transfer/death/clinic attendance. For example, all SMR records with a March 2019 date of discharge would be expected to be submitted to PHS by 12th May 2019. NHS Boards know how complete their SMR submissions are and the extent of any backlog (data which are not submitted by the 6-week deadline) and this information is used to produce completeness figures. Further details of SMR data completeness and timeliness can be found on the SMR Completeness and Timeliness web pages respectively.

Image showing a data flow diagram  illustrating the process of data collection and validation

Communication with data supply partners

The SMR01 and SMR04 data used in this publication is processed and monitored by the Data Support & Monitoring Team. More information can be found on the data monitoring page (external website).

Summary of data completeness

NHS Board
SMR01
SMR04
  Apr'21-Jun'21 Jul'21-Sep'21 Oct'21-Dec'21 Jan'22-Mar'22 Jul'21-Sep'21 Oct'21-Dec'21 Jan'22-Mar'22
Ayrshire & Arran 100% 100% 100% 100% 100% 100% 100%
Borders 100% 100% 100% 100% 100% 100% 100%
State Hospital         100% 100% 100%
Golden Jubilee 100% 100% 100% 100%      
Fife 99% 99% 99% 96% 100% 100% 100%
Greater Glasgow & Clyde 100% 99% 99% 99% 100% 100% 100%
Highland 99% 99% 99% 97% 95% 84% 78%
Lanarkshire

100%

100%

100%

99%

98%

99%

99%

Grampian

98%

98%

98%

97%

98%

99%

97%

Orkney

100%

100%

100%

100%

 

 

 

Lothian

98%

98%

97%

99%

100%

100%

100%

Tayside

99%

99%

99%

99%

99%

98%

97%

Forth Valley

98%

98%

98%

98%

100%

100%

100%

Western Isles

100%

100%

100%

100%

100%

100%

100%

Dumfries & Galloway

100%

99%

99%

99%

100%

100%

100%

Shetland

96%

94%

79%

93%

 

 

 

All NHS Boards

99%

99%

99%

99%

99%

99%

98%

The above table draws upon the data management page (external website) to present completeness for the patient-level datasets. The above completeness estimates were taken on 12 October 2022.

Data quality issues

General issues

For the financial year 2021/22, the SMR01 and SMR04 datasets were both estimated to be 99% complete (as at October 2022). Completeness varied between the NHS Boards, and further information on the completeness of individual NHS Boards can be found above.

NHS Highland data completeness for 2021/22 is slightly lower than other health boards, at 90%. This incompleteness particularly affects Quarter 3 (84%) and Quarter 4 (78%). A judgement has been made to proceed with the publication as the lower data completeness for this health board has only a very minor effect on the overall data quality for this release and is very unlikely to affect the patterns of activity seen or conclusions drawn at a national level or for NHS Highland. Data presented for the year 2021/22 is overall above the 90% PHS standard for recording of data.

Data quality assessment

The PHS Data Quality Assurance (DQA) team is responsible for ensuring that SMR datasets are accurate, consistent and comparable across time and between sources. The DQA team’s assessments web page (external website) contains details of past Data Quality Assurance Assessments of SMR01 and SMR04 data, including findings on the accuracy of submitted data items used in our analysis such as specialty and admission type. In order to assess the quality of recording, the DQA compares samples of records against source evidence in hospital systems and patient case notes. PHS have a 90% standard for accuracy and recording of data.

The Mental Health team who produce this publication also carry out thorough quality assurance (QA) checks on the data after extraction from the databases, in accordance with the PHS checking guidance (external website). For example, they compare figures for the same years between the current and previous publications in order to identify any large changes in the data sources. Additionally, they look at trends within the current publication in order to identify any unusual patterns. For changes or patterns in the data which cannot be explained by the known completeness estimates, the team contact the relevant board to highlight the issue and request an explanation. Any information provided by the board is included in this report.

Data sources

The data presented in this publication are sourced from hospital administrative systems across Scotland and extracted from the Scottish Morbidity Record (SMR) 04 and 01 datasets held by PHS. The specialty of learning disability is excluded from the figures published here.

Last updated: 21 March 2024
Was this page helpful?